Node electricity price prediction method and device, storage medium and terminal

Abstract

The invention provides a node electricity price prediction method and apparatus, a storage medium and a terminal. The method comprises the steps of obtaining a first feature data set and a second feature data set of a target node; inputting a first feature data set in the feature data set into a blocking recognition model to obtain a blocking recognition result; if the blocking identification result is non-blocking, inputting a second feature data set in the feature data set into a non-blocking scene electricity price prediction model to obtain an electricity price prediction result of the target node; if the blocking identification result is blocking, obtaining an electricity price prediction result of the target node by using a blocking scene electricity price prediction model; wherein the first feature data set comprises whole-network wind power output, whole-network photovoltaic output, urban wind speed and urban irradiation, and the second feature data set comprises a target node bidding space, a target node connecting line plan, target node wind power output, target node photovoltaic output and target node thermal power output. According to the invention, the precision of node electricity price prediction is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and relates to a node electricity price prediction method, and in particular to a node electricity price prediction method and device, a storage medium and a terminal. Background Art

[0002] The electricity spot market is one of the key mechanisms to guide energy investment and promote the consumption of new energy. At present, provinces such as Shanxi, Guangdong, Shandong, and Gansu have entered the formal operation stage of the electricity spot market, and provinces such as Western Inner Mongolia, Fujian, and Zhejiang are conducting long-term uninterrupted trial operations in the electricity spot market. Other provinces and regions are also actively developing rules and trial operations for the electricity spot market.

[0003] The domestic electricity spot market generally adopts node electricity prices as the market price mechanism. Node electricity prices can reflect the dual value of time and space. Node electricity prices are different at different times and locations. Strong price fluctuations have had a profound impact on energy investment and electricity trading. The traditional marketing model centered on electricity volume is no longer used, and the risks faced by energy companies in the electricity market have increased significantly.

[0004] However, there are many factors that affect node electricity prices, such as weather, market supply and demand, grid congestion, unit quotations, market price limits, etc. It is particularly difficult to predict node electricity prices in provinces with a high proportion of new energy or serious grid congestion. The node electricity price prediction accuracy of existing technologies is not high. How to accurately predict node electricity prices for a period of time in the future has become an urgent problem that the energy industry needs to solve. Summary of the invention

[0005] The object of the present invention is to provide a node electricity price prediction method and device, a storage medium and a terminal, which are used to solve the technical problem of poor node electricity price prediction accuracy in the prior art.

[0006] In a first aspect, the present invention provides a node electricity price prediction method, comprising:

[0007] Acquire a first feature data set and a second feature data set of a target node;

[0008] Inputting the first feature data set into a blocking recognition model to obtain a blocking recognition result;

[0009] If the blocking identification result is not blocked, the first feature data set is input into the unblocked scenario electricity price prediction model to obtain the electricity price prediction result of the target node; if the blocking identification result is blocked, the electricity price prediction result of the target node is obtained using the blocked scenario electricity price prediction model based on the second feature data set;

[0010] Among them, the first feature data set includes the total wind power output of the entire network, the total photovoltaic output of the entire network, urban wind speed and urban irradiation, and the second feature data set includes the target node bidding space, the target node interconnection line plan, the target node wind power output, the target node photovoltaic output and the target node thermal power output.

[0011] In one embodiment of the present invention, the congestion recognition model is a KNN clustering congestion recognition model.

[0012] In one embodiment of the present invention, the distance metric of the KNN clustering blockage identification model is the Minkowski distance.

[0013] In one embodiment of the present invention, obtaining the electricity price prediction result of the target node based on the second feature data set using the congestion scenario electricity price prediction model includes:

[0014] Acquire a node type of the target node based on the power flow position of the target node, wherein the node type includes a sending node and a receiving node;

[0015] Inputting the second feature data set into a sub-model corresponding to the node type of a blocking scenario electricity price prediction model to obtain an electricity price prediction result of the target node;

[0016] Among them, the sub-models of the blocking scenario electricity price prediction model include a sending-end blocking scenario electricity price prediction sub-model and a receiving-end blocking scenario electricity price prediction sub-model. The prediction factors of the sending-end blocking scenario electricity price prediction sub-model are the target node bidding space, the target node wind power output and the target node photovoltaic output. The prediction factors of the receiving-end blocking scenario electricity price prediction sub-model are the target node bidding space, the target node interconnection line plan and the target node thermal power output.

[0017] In one embodiment of the present invention, when the node type of the target node is a sending-end node, the second feature data set further includes sending-end wind speed and sending-end irradiance.

[0018] In one embodiment of the present invention, the electricity price prediction sub-model for the sending-end blocking scenario and the electricity price prediction sub-model for the receiving-end blocking scenario are both CNN-LSTM models.

[0019] In one embodiment of the present invention, the first feature data set also includes the network-wide coordinated load and tie line plan.

[0020] In a second aspect, the present invention further provides a node electricity price prediction device, characterized in that it comprises:

[0021] A data processing module, used to obtain a first feature data set and a second feature data set of a target node;

[0022] A congestion recognition module, configured to input the first feature data set into a congestion recognition model to obtain a congestion recognition result;

[0023] An electricity price prediction module, for inputting the second feature data set into an unblocked scenario electricity price prediction model to obtain an electricity price prediction result of the target node if the blocking identification result is unblocked; and for obtaining an electricity price prediction result of the target node using a blocked scenario electricity price prediction model if the blocking identification result is blocked;

[0024] Among them, the first feature data set includes the total wind power output of the entire network, the total photovoltaic output of the entire network, urban wind speed and urban irradiation, and the second feature data set includes the target node bidding space, the target node interconnection line plan, the target node wind power output, the target node photovoltaic output and the target node thermal power output.

[0025] In a third aspect, the present invention further provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the node electricity price prediction method as described above.

[0026] In a fourth aspect, the present invention further provides a terminal, comprising a processor and a memory, wherein the memory is communicatively connected to the processor;

[0027] The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the terminal executes the node electricity price prediction method as described above.

[0028] As described above, the node electricity price prediction method and device, storage medium and terminal of the present invention have the following beneficial effects:

[0029] 1. The present invention identifies the target node as blocked or unblocked through the blocking identification model, and selects the appropriate prediction model according to the identification result. This diversion strategy can effectively improve the accuracy of electricity price prediction, especially in complex grid blocking scenarios, and can accurately capture the fluctuations of the electricity market.

[0030] 2. Different characteristic data are input based on different node types (sending nodes and receiving nodes), and a dedicated sub-model is designed for each type of target node. This flexible model structure makes the prediction more in line with the node characteristics, further improving the accuracy and adaptability of electricity price prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A schematic flow chart of a node electricity price prediction method according to an embodiment of the present invention is shown.

[0032] Figure 2 A schematic diagram of the process of KNN clustering congestion identification in the node electricity price prediction method according to an embodiment of the present invention is shown.

[0033] Figure 3 A schematic diagram of the structure of a node electricity price prediction device according to an embodiment of the present invention is shown.

[0034] Figure 4 A schematic diagram of the structure of a terminal according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0035] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0036] The following will explain in detail the principles and implementation methods of the node electricity price prediction method and device, storage medium and terminal of this embodiment, so that those skilled in the art can understand the node electricity price prediction method and device, storage medium and terminal of this embodiment without creative work.

[0037] In order to solve the above technical problems existing in the prior art, an embodiment of the present invention provides a node electricity price prediction method.

[0038] Figure 1 A schematic diagram showing a flow chart of a node electricity price prediction method according to an embodiment of the present invention is shown. Figure 1 As shown, the node electricity price prediction method according to the embodiment of the present invention mainly includes steps S100 to S300.

[0039] Step S100: Acquire a first feature data set and a second feature data set of a target node.

[0040] Specifically, the target node is the node whose electricity price is to be predicted. The node electricity price prediction of the embodiment of the present invention needs to obtain two types of feature data sets: a first feature data set and a second feature data set, which will provide the required input for the subsequent electricity price prediction model. Among them, the first feature data set includes the wind power output of the whole network, the photovoltaic output of the whole network, the urban wind speed and the urban irradiation; the second feature data set includes the target node bidding space, the target node interconnection line plan, the target node wind power output, the target node photovoltaic output and the target node thermal power output.

[0041] The first characteristic data set characterizes external factors that affect electricity price fluctuations. The total grid wind power output is the wind power generation capacity within the entire grid, expressed in megawatts (MW). The output of wind power is an important supply factor in the electricity market. An increase in wind power output usually leads to lower electricity prices. The total grid photovoltaic output is the photovoltaic power generation within the entire grid, expressed in megawatts (MW). Photovoltaic power generation is affected by factors such as sunshine intensity, seasonal changes, and meteorological conditions. Urban wind speed is the wind speed level in the urban area, which affects the operation of wind power and other energy facilities. Wind speed data can indirectly affect wind power output, and then affect the overall power supply of the power grid. Urban irradiation is a measurement of solar radiation intensity, which affects the efficiency of photovoltaic power generation. Irradiation data also has a direct impact on the supply of the power grid.

[0042] The second feature data set characterizes the specific data related to the target node (i.e., a specific power production facility or region), which directly affects the power production and market bidding space of the node. The target node bidding space is the bidding space or potential adjustment room of the target node in the power market, that is, the supply and demand gap, bidding flexibility and other factors of the node in the power market. The target node interconnection line plan is a plan that describes the interconnection line plan of the power grid, that is, the power transmission channel and its planning between the target node and other nodes. The interconnection line plan determines the power exchange capacity between the target node and other regions, which may affect the efficiency of power price transmission and regulation. The target node wind power output is the target node's own wind power generation capacity or wind power output. The target node photovoltaic output is the target node's own photovoltaic power generation capacity. The target node thermal power output is the power generation of the target node's thermal power facilities.

[0043] The first feature data set provides the meteorological and energy production data of the entire network, which can reflect the large-scale external environment and energy supply situation; the second feature data set provides the target node's own power production capacity and market bidding ability, which specifically affects the electricity price forecast of the node. By inputting these two types of feature data sets into the subsequent model, comprehensive basic data support can be provided for electricity price forecasting.

[0044] Step S200: inputting the first feature data set into a blockage recognition model to obtain a blockage recognition result.

[0045] Specifically, step S200 of the embodiment of the present invention uses a congestion identification model to detect whether a target node is congested, and the congestion identification result will directly determine the model path selection for subsequent electricity price prediction. The data input to the congestion identification model is a first feature data set, specifically including the wind power output of the entire network, the photovoltaic output of the entire network, the urban wind speed and the urban irradiation. These feature data jointly reflect the power supply capacity and external environmental conditions of the power grid, and are important bases for judging congestion.

[0046] The congestion identification model is a deep learning neural network model used to detect whether the grid operation status is congested. It is based on the first feature data set of the target node for identification to determine whether the target node is in a congested state. The output of the congestion identification model is congestion or non-congestion. Congestion (Congestion) means that there is a bottleneck in the operation of the grid and electricity cannot be transmitted smoothly in some areas. No congestion (No Congestion) means that the grid is operating normally and there is no transmission bottleneck. The congestion identification result is the diversion point of the subsequent electricity price prediction process, which determines the selection of the subsequent prediction model. If there is no congestion, the grid transmission is normal, and the node electricity price prediction is based on the non-congestion scenario electricity price prediction model; if there is congestion, the grid has a bottleneck, and the node electricity price is significantly affected by the local supply and demand imbalance. The congestion scenario electricity price prediction model is used to capture the specific impact of congestion on electricity prices.

[0047] The method for obtaining the blocking recognition model is as follows: the preset artificial intelligence model is trained with a blocking recognition data set, wherein the blocking recognition data set includes multiple blocking recognition data, each of which uses all the features in the first feature data set as prediction factors, and the blocking recognition result as the prediction result. It should be noted that for each piece of data in the blocking recognition data set, the data of the blocking reference node that is sensitive to blocking must be collected, and at the same time, the number of positive samples (i.e., the sample label is a blocked node) and the number of negative samples (i.e., the sample label is an unblocked node) in the blocking recognition data set are balanced.

[0048] Therefore, step S200 can quickly determine whether the grid operation state is blocked by inputting the first feature data set into the congestion identification model. The congestion identification result is an important diversion point in the electricity price prediction process, which helps to select the most appropriate prediction model according to the actual situation to improve the prediction accuracy.

[0049] In a specific implementation of this embodiment, the preset artificial intelligence model is a KNN (K-Nearest Neighbors) model, that is, the congestion identification model is a KNN clustering congestion identification model. The congestion identification model is a congestion identification model based on the KNN clustering algorithm. By using the KNN algorithm, it is possible to identify based on the first feature data set of the target node to determine whether there is a congestion in the operating status of the power grid. The KNN clustering model classifies by comparing the similarity (distance) between the sample points and the labeled samples in the training set, and uses the distribution characteristics of known data points to determine the category of the new input data points without the need for an explicit training process. It is suitable for small sample classification problems and can flexibly adapt to the multi-dimensional input of feature data. In particular, when the first feature data set contains multiple complex features (such as wind speed, irradiation, etc.), KNN can handle the nonlinear relationship between these data. Specifically, Figure 2 FIG. 1 shows a schematic diagram of the process of KNN clustering congestion identification in the node electricity price prediction method of an embodiment of the present invention, referring to FIG. Figure 2 As shown, using the KNN clustering congestion recognition model to obtain the congestion recognition result of the target node includes steps S201-S204:

[0050] Step S201: Acquire a first feature data set of a target node as input data of a KNN clustering blocking recognition model;

[0051] Step S202: calculating the distance between the target node and the labeled samples in the KNN clustering blockage recognition model training set based on the first feature data set to obtain a distance calculation result;

[0052] Step S203: based on the distance calculation result, find the k samples closest to the target node;

[0053] Specifically, the distances are sorted from small to large, and the k neighbors closest to the target node are selected. K represents the number of the nearest neighbors, and an appropriate k can be set according to actual needs.

[0054] Step S204: according to the labels of the k samples (ie, blocked or unblocked), a preset classification method is used to obtain a blocking identification result corresponding to the target node.

[0055] The preset classification method includes a majority voting method and a weighted method. Taking the majority voting method as an example, if the majority of the k samples are "blocked", the blocking identification result of the target node is blocked. Otherwise, the blocking identification result of the target node is not blocked. Based on the above description, those skilled in the art can implement step S204 by the weighted method.

[0056] By adopting the KNN clustering congestion identification model, the congestion identification process becomes more intuitive and efficient. By utilizing the similarity relationship between historical sample data and real-time feature data, the model can flexibly judge the congestion status of the power grid and provide accurate input basis for subsequent electricity price prediction.

[0057] Optionally, the distance metric of the KNN clustering blockage identification model is the Minkowski distance. Specifically, the calculation method of the Minkowski distance refers to the following formula:

[0058]

[0059] Among them, d(xi,xj) represents the Minkowski distance result between xj and xi, m represents the mth feature of the sample, and r represents the parameter that controls the distance calculation method. Optionally, r is 2. In the KNN clustering blocking recognition model, Minkowski distance is used as the distance measurement method. Its flexible parameter adjustment mechanism enables the model to adapt to multi-dimensional complex data and ensure the robustness of recognition.

[0060] Step S300: If the blocking identification result is not blocked, the first feature data set is input into the unblocked scenario electricity price prediction model to obtain the electricity price prediction result of the target node; if the blocking identification result is blocked, the blocking scenario electricity price prediction model is used based on the second feature data set to obtain the electricity price prediction result of the target node.

[0061] Specifically, in step S300, according to the different blocking identification results, the branch of the electricity price prediction model corresponding to the blocking identification result is selected, and the branch of the electricity price prediction model includes an unblocked scenario electricity price prediction model and a blocked scenario electricity price prediction model. If the output of the blocking identification model is unblocked, it indicates that the power grid connection state of the target node is good, the data transmission is smooth, and the power grid operation is in a normal state. At this time, the first feature data set is input into the unblocked scenario electricity price prediction model to predict the electricity price of the target node. The unblocked scenario electricity price prediction model focuses on the macro characteristics of the entire network (the first feature data set), pays attention to the electricity price fluctuation trend of the entire network, and makes the prediction result more accurate. At the same time, the complex data input of the local node is reduced, and the model calculation cost is reduced. If the output of the blocking identification model is blocked, it indicates that the power grid connection of the target node is blocked, and there may be a transmission bottleneck, an imbalance between power supply and demand, or other anomalies. At this time, a special blocking scenario electricity price prediction model is used to predict the electricity price of the target node. In the blocked state, the dynamic changes of the local features of the target node have a more significant impact on the electricity price, and the features in the second feature data set are local features of the target node. Therefore, the second feature data set is selected to be input into the blocked scenario electricity price prediction model for electricity price prediction. For the complex and fluctuating electricity prices in the blocked state, the blocked scenario electricity price prediction model is more suitable for the target node in the blocked state and can accurately predict the results in unstable conditions.

[0062] Therefore, the present invention dynamically switches between the non-blocking scenario electricity price prediction model and the blocking scenario electricity price prediction model, calls different feature data sets as input data, improves the accuracy of the prediction, and reasonably avoids the waste of calculation caused by information redundancy in the entire network.

[0063] Optionally, obtaining the electricity price prediction result of the target node using the blocking scenario electricity price prediction model based on the second feature data set includes: obtaining the node type of the target node based on the flow position of the target node, the node type includes a sending node and a receiving node; inputting the second feature data set into the sub-model corresponding to the node type of the blocking scenario electricity price prediction model to obtain the electricity price prediction result of the target node; wherein the sub-model of the blocking scenario electricity price prediction model includes a sending-end blocking scenario electricity price prediction sub-model and a receiving-end blocking scenario electricity price prediction sub-model, the prediction factors of the sending-end blocking scenario electricity price prediction sub-model are the target node bidding space, the target node wind power output and the target node photovoltaic output, and the prediction factors of the receiving-end blocking scenario electricity price prediction sub-model are the target node bidding space, the target node tie line plan and the target node thermal power output. Specifically, in the embodiment of the present invention, under the blocking situation, the blocking scenario electricity price prediction model is further refined to select the corresponding sub-model according to the type of node. A sending node means that the target node is located at the source of power output and is responsible for transmitting electric energy to the power grid or the receiving area; a receiving node means that the target node is located at the end point of power reception and is responsible for obtaining electric energy from the power grid to meet local demand. Technical personnel in this field can determine that the target node is a receiving node or a sending node based on the flow position.

[0064] The electricity price prediction submodel for the sending-end blocking scenario is used to predict the electricity price when the target node is the sending-end node. Its prediction factors are the bidding space of the target node, the wind power output of the target node, and the photovoltaic output of the target node. Since the target node is the sending-end node at this time, the prediction factors of the electricity price prediction submodel for the sending-end blocking scenario are also the bidding space of the sending-end node, the wind power output of the sending-end node, and the photovoltaic output of the sending-end node. The bidding space of the sending-end node represents the market competition degree of the sending-end node, the wind power output of the sending-end node represents the wind power generation situation in the region, and the photovoltaic output of the sending-end node represents the photovoltaic power generation situation in the region. The electricity price prediction submodel for the sending-end blocking scenario focuses on the renewable energy generation characteristics of the sending-end node, and makes predictions based on the characteristics of the bidding space and output characteristics to obtain the electricity price prediction results of the sending-end node. The method of obtaining the sending-end blocking scenario electricity price prediction sub-model is as follows: the preset neural network model is trained with the sending-end blocking electricity price prediction data set. The sending-end blocking electricity price prediction data set includes multiple sending-end blocking prediction data. Each sending-end blocking prediction data uses the target node bidding space, the target node wind power output and the target node photovoltaic output as prediction factors, and the sending-end blocking prediction result is used as the prediction result.

[0065] The receiving-end blocking scenario electricity price prediction submodel is used to predict the electricity price when the target node is the receiving node. Its prediction factors are the target node bidding space, the target node tie line plan and the target node thermal power output. Since the target node is the receiving node at this time, the prediction factors of the receiving-end blocking scenario electricity price prediction submodel are also the receiving node bidding space, the receiving node tie line plan and the receiving node thermal power output. The receiving node bidding space represents the market electricity price competitiveness of the receiving node, the receiving node tie line plan represents the execution of the plan for the receiving node to obtain electricity from the tie line under blocking conditions, and the receiving node thermal power output represents the thermal power generation. The receiving-end blocking scenario electricity price prediction submodel focuses on the tie line characteristics and thermal power stability of the receiving node, and predicts the receiving node electricity price based on the bidding space, tie line plan and thermal power output. The method of obtaining the receiving-end blocking scenario electricity price prediction sub-model is as follows: the preset neural network model is trained with the receiving-end blocking electricity price prediction data set. The receiving-end blocking electricity price prediction data set includes multiple receiving-end blocking prediction data. Each receiving-end blocking prediction data uses the target node bidding space, the target node interconnection line plan, and the target node thermal power output as prediction factors, and the receiving-end blocking prediction result is used as the prediction result.

[0066] In the above implementation of step S300, two sub-models are used as the electricity price prediction model for the blocking scenario, and special prediction factors are designed according to the different characteristics of the sending node and the receiving node, which enhances the model adaptability and improves the accuracy of electricity price prediction.

[0067] In addition, in another embodiment of step S300, the congestion scenario electricity price prediction model is an end-to-end model without further subdividing the node type of the target node. The congestion scenario electricity price prediction model uses all features in the second feature data set as prediction factors and the electricity price prediction result as the prediction result.

[0068] Optionally, when the node type of the target node is a sending node, the second feature data set also includes sending wind speed and sending irradiation. Wind speed and irradiation are direct factors affecting wind power and photovoltaic output. Introducing these two parameters into the second feature data set helps to improve the prediction accuracy of the electricity price of the sending node.

[0069] Optionally, the preset neural network model is a CNN-LSTM model, that is, the electricity price prediction sub-model for the sending-end blocking scenario and the electricity price prediction sub-model for the receiving-end blocking scenario are both CNN-LSTM models. The CNN-LSTM model is composed of a convolutional neural network (CNN) and a long short-term memory network (LSTM), wherein the convolutional neural network generally includes modules such as an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The extraction of data features mainly depends on the convolution kernel in the convolutional layer, and the calculation process is as follows:

[0070]

[0071] In the formula, is the output value of the i-th eigenvalue matrix of the k-1th layer, is the convolution kernel between the jth eigenvalue matrix of the kth layer and the ith eigenvalue matrix of the k-1th layer, is the bias of the j-th eigenvalue matrix of the k-th layer, C j is the set of input eigenvalue matrices, is the output value of the j-th eigenvalue matrix of the k-th layer, * represents the convolution calculation, and f(g) is the activation function, which is usually the RELU function or the Sigmoid function.

[0072] The long short-term memory network can effectively solve the gradient vanishing problem in the traditional model training process. It is mainly composed of input gate, forget gate, and output gate, and updates the state at the moment according to the following rules:

[0073] g i =σ(W i [h t-1 x t ]+b i )

[0074] g f =σ(W f [h t-1 x t ]+b f )

[0075] g o =σ(W o [h t-1 x t ]+b o )

[0076]

[0077] h t =g o *tanhc t

[0078] In the formula, x t is the input vector at the time instant, h t is the hidden state vector at the moment, c t is the memory state vector at the moment, g i , g f , g o Represent the input gate, forget gate and output gate respectively, W i , W f , W o , W c Represents the weight matrix of each connection layer, b i 、b f、b o 、b c They represent the bias terms of each connection layer, and σ is the Sigmoid function.

[0079] The embodiment of the present invention utilizes the CNN-LSTM model to give full play to the advantages of convolutional neural networks (CNN) and long short-term memory networks (LSTM), and combines the capabilities of spatial feature extraction and time series modeling. Therefore, it can better handle high-dimensional and dynamically changing features and improve the accuracy of electricity price prediction.

[0080] Optionally, the first feature data set also includes the network-wide unified load and interconnection line plan. The network-wide unified load is the total power demand of the power grid at a certain moment, reflecting the real-time power demand of the entire power grid. The interconnection line plan is the power dispatching plan between cross-regional power grids, reflecting the allocation of power between different regions. By adding the network-wide unified load and interconnection line plan to the first feature data set, global support is added to the congestion identification model and the non-congestion scenario electricity price prediction model, so that the model can more comprehensively adapt to the dynamic and complex power grid operation status, thereby generating more accurate prediction results.

[0081] The protection scope of the node electricity price prediction method of the embodiment of the present invention is not limited to the execution order of the steps listed in this embodiment. All solutions implemented by adding, reducing or replacing steps in the prior art based on the principles of the present invention are included in the protection scope of the present invention.

[0082] The node electricity price prediction method of the embodiment of the present invention identifies the target node as blocked or unblocked through a blocking identification model, and selects a suitable prediction model based on the identification result. This diversion strategy can effectively improve the accuracy of electricity price prediction, especially in complex power grid blocking scenarios, and can accurately capture fluctuations in the electricity market. Different feature data are input based on different node types (sending nodes and receiving nodes), and a dedicated sub-model is designed for each type of target node, which further improves the accuracy and adaptability of electricity price prediction.

[0083] In order to solve the above technical problems existing in the prior art, an embodiment of the present invention further provides a node electricity price prediction device.

[0084] Figure 3 The schematic diagram of the node electricity price prediction device according to the embodiment of the present invention is shown. Figure 3 As shown, the node electricity price prediction device of the embodiment of the present invention includes a data processing module, a congestion identification module, and an electricity price prediction module;

[0085] A data processing module, used to obtain a first feature data set and a second feature data set of a target node;

[0086] A congestion recognition module, configured to input a first feature data set in the feature data set into a congestion recognition model to obtain a congestion recognition result;

[0087] An electricity price prediction module, for inputting a second feature data set in the feature data set into an unblocked scenario electricity price prediction model to obtain an electricity price prediction result of the target node if the blocking identification result is unblocked; and for obtaining an electricity price prediction result of the target node using a blocked scenario electricity price prediction model if the blocking identification result is blocked;

[0088] Among them, the first feature data set includes the wind power output of the entire network, the photovoltaic output of the entire network, the urban wind speed, and the urban irradiation; the second feature data set includes the target node bidding space, the target node interconnection line plan, the target node wind power output, the target node photovoltaic output, and the target node thermal power output.

[0089] The node electricity price prediction device of the embodiment of the present invention identifies the target node as blocked or unblocked through the blocking recognition model, and selects the appropriate prediction model according to the recognition result. This diversion strategy can effectively improve the accuracy of electricity price prediction, especially in complex power grid blocking scenarios, and can accurately capture the fluctuations in the electricity market. Different feature data are input based on different node types (sending nodes and receiving nodes), and a dedicated sub-model is designed for each type of target node, which further improves the accuracy and adaptability of electricity price prediction.

[0090] In order to solve the above technical problems existing in the prior art, an embodiment of the present invention further provides a storage medium on which a computer program is stored. When the program is executed by a processor, all steps of the node electricity price prediction method of the above embodiment are implemented.

[0091] The specific steps of the node electricity price prediction method and the beneficial effects obtained by applying the readable storage medium provided by the embodiment of the present invention are the same as those in the above embodiment and will not be described in detail here.

[0092] A person of ordinary skill in the art can understand that all or part of the steps in the method for implementing the above-mentioned embodiment can be completed by instructing a processor through a program, and the program can be stored in a computer-readable storage medium, and the storage medium is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state hard disk, a magnetic tape, a floppy disk, an optical disc, and any combination thereof. The above-mentioned storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a digital video disc (DVD)), or a semiconductor medium (for example, a solid-state disk (SSD)), etc.

[0093] In order to solve the above technical problems existing in the prior art, an embodiment of the present invention further provides a terminal. Figure 4 A schematic diagram of the structure of the terminal according to an embodiment of the present invention is shown, referring to Figure 4 As shown, the terminal of the embodiment of the present invention includes a processor and a memory, and the memory and the processor are communicatively connected; the memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory, so that the terminal executes all steps of the node electricity price prediction method of the above embodiment.

[0094] The specific steps of the node electricity price prediction method and the beneficial effects obtained by applying the terminal provided by the embodiment of the present invention are the same as those in the above embodiment and will not be described in detail here.

[0095] It should be noted that the memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. Similarly, the processor may also be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0096] Although the embodiments disclosed in the present invention are as above, the contents described are only embodiments adopted for facilitating the understanding of the present invention and are not intended to limit the present invention. Any technician in the technical field to which the present invention belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present invention, but the protection scope of the present invention shall still be subject to the scope defined in the attached claims.

Claims

1. A node electricity price prediction method, comprising: Acquire a first feature data set and a second feature data set of a target node; Inputting the first feature data set into a blocking recognition model to obtain a blocking recognition result; If the blocking identification result is not blocked, the first feature data set is input into the unblocked scenario electricity price prediction model to obtain the electricity price prediction result of the target node; if the blocking identification result is blocked, the electricity price prediction result of the target node is obtained using the blocked scenario electricity price prediction model based on the second feature data set; Among them, the first feature data set includes the total wind power output of the entire network, the total photovoltaic output of the entire network, urban wind speed and urban irradiation, and the second feature data set includes the target node bidding space, the target node interconnection line plan, the target node wind power output, the target node photovoltaic output and the target node thermal power output.

2. The node electricity price prediction method according to claim 1, characterized in that: The congestion recognition model is a KNN clustering congestion recognition model.

3. The node electricity price prediction method according to claim 2, characterized in that: The distance metric of the KNN clustering blockage identification model is the Minkowski distance.

4. The node electricity price prediction method according to claim 1, characterized in that: Acquiring the electricity price prediction result of the target node based on the second feature data set by using the congestion scenario electricity price prediction model includes: Acquire a node type of the target node based on the power flow position of the target node, wherein the node type includes a sending node and a receiving node; Inputting the second feature data set into a sub-model corresponding to the node type of a blocking scenario electricity price prediction model to obtain an electricity price prediction result of the target node; Among them, the sub-models of the blocking scenario electricity price prediction model include a sending-end blocking scenario electricity price prediction sub-model and a receiving-end blocking scenario electricity price prediction sub-model. The prediction factors of the sending-end blocking scenario electricity price prediction sub-model are the target node bidding space, the target node wind power output and the target node photovoltaic output. The prediction factors of the receiving-end blocking scenario electricity price prediction sub-model are the target node bidding space, the target node interconnection line plan and the target node thermal power output.

5. The node electricity price prediction method according to claim 4, characterized in that: When the node type of the target node is a sending-end node, the second feature data set further includes sending-end wind speed and sending-end irradiance.

6. The node electricity price prediction method according to claim 4, characterized in that: The electricity price prediction sub-model for the sending-end blocking scenario and the electricity price prediction sub-model for the receiving-end blocking scenario are both CNN-LSTM models.

7. The node electricity price prediction method according to claim 1, characterized in that: The first characteristic data set also includes the overall network load and interconnection line plan.

8. A node electricity price prediction device, comprising: A data processing module, used to obtain a first feature data set and a second feature data set of a target node; A congestion recognition module, configured to input the first feature data set into a congestion recognition model to obtain a congestion recognition result; An electricity price prediction module, configured to input the first feature data set into an unblocked scenario electricity price prediction model to obtain an electricity price prediction result of the target node if the blocking identification result is unblocked; and to obtain an electricity price prediction result of the target node using a blocked scenario electricity price prediction model based on the second feature data set if the blocking identification result is blocked; Among them, the first feature data set includes the total wind power output of the entire network, the total photovoltaic output of the entire network, urban wind speed and urban irradiation, and the second feature data set includes the target node bidding space, the target node interconnection line plan, the target node wind power output, the target node photovoltaic output and the target node thermal power output.

9. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the node electricity price prediction method according to any one of claims 1 to 7 is implemented.

10. A terminal, characterized in that: It includes a processor and a memory, and the memory is communicatively connected to the processor; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the terminal executes the node electricity price prediction method as described in any one of claims 1 to 7.

Images