Method and device for determining electricity price prediction model and electricity price prediction method

Abstract

The invention discloses an electricity price prediction model determination method and device, an electricity price prediction method, a storage medium, an electronic device and a computer program product. The electricity price prediction model determination method comprises the steps of obtaining a first historical data set of at least one station in a preset range of a target station; constructing a rank set pair between the second historical data set of the target station and the first historical data set to determine a target historical data set having the highest similarity with the second historical data of the target station; and training a pre-constructed neural network model by using the target historical data set as a training sample to determine an electricity price prediction model of the target station. According to the method, the data accumulated by the stations around the target station is analyzed to obtain the historical data most conforming to the condition of the target station, and the electricity price prediction model obtained by training the target historical data set can accurately predict the electricity price condition of the target station.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of electricity price prediction, and specifically, to a method and device for determining an electricity price prediction model, an electricity price prediction method, a storage medium, an electronic device, and a computer program product. Background Art

[0002] With the rapid development of the new energy industry, large-scale new energy power stations are gradually being integrated into the power grid. Since the marginal cost of new energy such as wind energy and solar energy is low, and electricity cannot be stored in large quantities, it is necessary to ensure that the supply and demand are balanced as much as possible. However, the randomness and instability of new energy such as wind energy and solar energy make the day-ahead electricity price fluctuate strongly, showing non-stationary characteristics. In the context of a power market with a high proportion of new energy, establishing a high-accuracy day-ahead electricity price prediction model is of great significance for balancing the supply and demand relationship and saving energy.

[0003] In related technologies, statistics, machine learning and other methods are usually used to predict electricity prices.

[0004] However, due to the strong randomness and volatility of renewable energy, the prediction accuracy of statistical methods is poor. The machine learning method is more complicated and has poor generalization ability, so the prediction accuracy in new scenarios is poor. Summary of the invention

[0005] Embodiments of the present application provide a method and device for determining an electricity price prediction model, an electricity price prediction method, a storage medium, an electronic device, and a computer program product.

[0006] According to one aspect of an embodiment of the present application, a method for determining an electricity price prediction model is provided, the method comprising: obtaining a first historical data set of at least one station within a preset range of a target station, wherein the first historical data set comprises a one-to-one corresponding time series of power data and environmental data, and the power data comprises at least electricity prices; constructing a ranked set pair between a second historical data set of the target station and the first historical data set to determine a target historical data set having the highest similarity to the second historical data of the target station; using the target historical data set as a training sample to train a pre-constructed neural network model to determine the electricity price prediction model of the target station.

[0007] In an exemplary embodiment, the method of constructing a ranked set pair between the second historical data set of the target station and the first historical data set to determine the target historical data set with the highest similarity to the second historical data of the target station includes: sorting the elements of each attribute in the first historical data set according to a preset rank to obtain multiple first data ranked sets of the first historical data set, wherein different first data ranked sets contain elements with different attributes; sorting the elements of each attribute in the second historical data set according to the preset rank to obtain multiple second data ranked sets of the second historical data set; using a preset evaluation strategy to respectively determine the degree of correlation between the multiple first data ranked sets with the same attributes and the multiple second data ranked sets; determining a group of first data ranked sets with the highest degree of correlation with the multiple second data ranked sets as the target historical data set with the highest similarity to the second historical data of the target station.

[0008] In an exemplary embodiment, before using the target historical data set as a training sample to train a pre-constructed neural network model to determine the electricity price prediction model of the target site, the determination of the electricity price prediction model also includes: constructing an initial neural network model; using the first historical data set to train the initial neural network model to obtain the pre-constructed neural network model, wherein the pre-constructed neural network model includes a likelihood function.

[0009] In an exemplary embodiment, the target historical data set is used as a training sample to train a pre-constructed neural network model to determine the electricity price prediction model of the target site, including: adding a fully connected layer on the basis of the pre-constructed neural network model to obtain a first electricity price prediction model; while the parameters of the pre-constructed neural network model remain unchanged, the target historical data set is used to train the first electricity price prediction model, and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site.

[0010] In an exemplary embodiment, while the parameters of the pre-constructed neural network model remain unchanged, the target historical data set is used to train the first electricity price prediction model, and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site, including: while the parameters of the pre-constructed neural network model remain unchanged, the target historical data set is used to train the first electricity price prediction model, and the parameters of the fully connected layer are adjusted to obtain the second electricity price prediction model; the target historical data set is used to train the second electricity price model, and the parameters of the pre-constructed neural network model and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site, wherein the adjustment range of the parameters of the pre-constructed neural network model is smaller than the preset range.

[0011] According to another aspect of an embodiment of the present application, a method for predicting electricity prices is also provided, which uses an electricity price prediction model determined by the method for determining an electricity price prediction model as described above, and the method includes: obtaining a data set of a target site for a day to be predicted; and applying the electricity price prediction model to determine a time series of the electricity price for the day to be predicted.

[0012] According to another aspect of an embodiment of the present application, a device for determining an electricity price prediction model is also provided, wherein the device comprises:

[0013] A data acquisition module, configured to acquire a first historical data set of at least one station within a preset range of the target station, wherein the first historical data set includes a time series of power data and environmental data corresponding to each other, and the power data includes at least power price;

[0014] A rank construction module, used to construct a rank set pair between the second historical data set of the target station and the first historical data set, so as to determine a target historical data set with the highest similarity to the second historical data of the target station;

[0015] The training module is used to use the target historical data set as a training sample to train a pre-built neural network model to determine the electricity price prediction model of the target site.

[0016] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining the electricity price prediction model when running.

[0017] According to another aspect of an embodiment of the present application, there is also provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the method for determining the electricity price prediction model through the computer program.

[0018] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps of the method described in each embodiment of the present application are implemented.

[0019] In the embodiment of the present application, the method for determining the above-mentioned electricity price prediction model first obtains the first historical data set of at least one station within the preset range of the target station. Since the power generation conditions and the influencing factors of electricity price fluctuations of the new energy stations located closer are roughly the same, the historical data of the stations within the preset range of the target station can be used as a reference for the data of the target station. Then, a rank set pair is constructed between the second historical data set and the first historical data set of the target station, so that it is convenient to determine the target historical data set with the highest similarity to the second historical data of the target station through rank set pair analysis, and obtain the target historical data set that can best represent the historical data of the target station, which is equivalent to the data accumulation of the target station. The target historical data set is used as a training sample to train the pre-constructed neural network model, and the electricity price prediction model of the target station can be determined, so that the data accumulated by the stations around the target station are analyzed to obtain the historical data that best meets the situation of the target station, and then the electricity price prediction model obtained by training the target historical data set can accurately predict the electricity price situation of the target station. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0022] Figure 1 It is a hardware structure block diagram of the method for determining the electricity price prediction model of the embodiment of the present application;

[0023] Figure 2 is a flow chart of a method for determining an electricity price prediction model according to an embodiment of the present application;

[0024] Figure 3 is a flow chart of a method for determining a target historical data set according to an embodiment of the present application;

[0025] Figure 4 is a flow chart of a method for training a neural network according to an embodiment of the present application;

[0026] Figure 5 is a schematic diagram of a method for training a neural network according to an embodiment of the present application;

[0027] Figure 6 is a flow chart of a method for determining an electricity price prediction model according to an embodiment of the present application;

[0028] Figure 7 This is a second flowchart of a method for determining an electricity price prediction model according to an embodiment of the present application;

[0029] Figure 8 is a flow chart of a method for predicting electricity price according to an embodiment of the present application;

[0030] Fig. 9 It is a structural block diagram of an optional device for determining an electricity price prediction model according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0033] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or a similar computing device. Taking running on a computer terminal as an example, Figure 1 1 is a hardware structure block diagram of a computer terminal for determining a power price prediction model according to an embodiment of the present application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1Only one is shown in the figure) processor 102 (processor 102 may include but is not limited to a microprocessor (Microprocessor Unit, referred to as MPU) or a programmable logic device (Programmable logic device, referred to as PLD)) and a memory 104 for storing data. In an exemplary embodiment, the computer terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 A different configuration with more features is shown.

[0034] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining the electricity price prediction model in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0035] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] In this embodiment, a method for determining an electricity price prediction model is provided. Figure 2 4 is a flowchart of an optional method for determining an electricity price prediction model according to an embodiment of the present application, the process includes the following steps S200-S220:

[0037] Step S200: Acquire a first historical data set of at least one station within a preset range of a target station.

[0038] Among them, the first historical data set includes a one-to-one corresponding time series of power data and environmental data, and the power data includes at least electricity prices, which can be specifically the historical day-ahead electricity prices. Power data can also include quotations of photovoltaic power generation units, loads of photovoltaic power generation units, wind power generation, photovoltaic power generation, nuclear power generation, operation modes of thermal power units (including operating capacity, reduced output, etc.), disclosure data (including direct load regulation, inter-provincial interconnection line power reception) and other data. Environmental data can include historical weather, including irradiance, temperature, humidity, wind speed and other data. The power data and environmental data are recorded in units of days (24 hours), that is, the power data and environmental data of the same day are corresponding, or recorded in units of preset hours (for example, 12 hours), and the first historical data set can be obtained.

[0039] Among them, the stations within the preset range of the target station are the stations that are closer to the target station. Because the power generation conditions and factors affecting electricity price fluctuations of the stations that are closer are roughly the same, the data of the stations within the preset range of the target station can be used as a reference. Secondly, the stations within the preset range of the target station specifically refer to stations that have been established longer than the target station. Because the stations that were built and joined the market earlier have accumulated more data than the target station, they have a larger amount of data accumulation, so the training data is more sufficient.

[0040] Step S210: construct a rank set pair between the second historical data set and the first historical data set of the target station to determine a target historical data set having the highest similarity to the second historical data of the target station.

[0041] Among them, the historical environmental data of the target station can be obtained through the Meteorological Bureau, and the disclosed load data and node electricity price data of the target station can also be queried to obtain data, thereby obtaining a second historical data set of the target station.

[0042] Among them, set pair analysis is an analytical method to find the identity, difference and opposition between two things, while rank set pair analysis is an analytical method applied to time series prediction. This is a type of statistical analysis technology based on data sorting rather than specific values. Its core idea is to convert raw data into ranks and make inferences and analyses by comparing the sizes of ranks. The rank set pair between the second historical data set and the first historical data set is to rank both the second historical data set and the first historical data set, and then the sorted second historical data set and the first historical data set are each other's rank set pairs.

[0043] Step S220, using the target historical data set as a training sample, training the pre-built neural network model to determine the electricity price prediction model of the target site.

[0044] Among them, the target station in this application is defined as a newly established station with very little data accumulation, so the amount of data accumulated by itself is insufficient. If only its own accumulated data is used for training in the process of training electricity price prediction, it is easy to cause overfitting of model training. The target historical data set is the data accumulated by the stations within the preset range of the target station, and the data volume is sufficient. And the target historical data set can be equivalent to the data accumulated by the target station. Therefore, the problem of insufficient data accumulation of the target station can be solved by using the target historical data set for training.

[0045] In this embodiment, the first historical data set of at least one station within the preset range of the target station is first obtained. Since the power generation conditions and the factors affecting the fluctuation of electricity prices of the new energy stations located closer are roughly the same, the historical data of the stations within the preset range of the target station can be used as a reference for the data of the target station. Then, a rank set pair is constructed between the second historical data set of the target station and the first historical data set, so that it is convenient to determine the target historical data set with the highest similarity to the second historical data of the target station through rank set pair analysis, and obtain the target historical data set that can best represent the historical data of the target station, which is equivalent to the data accumulation of the target station. The target historical data set is used as a training sample to train the pre-constructed neural network model, and the electricity price prediction model of the target station can be determined, so that the data accumulated by the stations around the target station are analyzed to obtain the historical data that best meets the situation of the target station, and then the electricity price prediction model obtained by training the target historical data set can accurately predict the electricity price situation of the target station.

[0046] In one embodiment, Figure 3 As shown, step S210 is to construct a rank set pair between the second historical data set of the target station and the first historical data set to determine the target historical data set with the highest similarity to the second historical data of the target station. It includes: steps S300-S330.

[0047] Step S300: sort the elements of each attribute in the first historical data set according to a preset rank, to obtain a plurality of first data rank sets of the first historical data set.

[0048] Among them, different first data ranking sets contain elements with different attributes. Attributes are various parameters of power stations, such as electricity prices, photovoltaic generator quotation, photovoltaic generator load, wind power generation, photovoltaic power generation, nuclear power generation, etc. The data ranking set includes time series data of an attribute, and each attribute is sorted according to a preset rank, for example, from large to small.

[0049] Step S310: sort the elements of each attribute in the second historical data set according to a preset rank to obtain a plurality of second data rank sets of the second historical data set.

[0050] The elements of each attribute in the second historical data set are also sorted according to a preset rank, so as to facilitate subsequent comparison of multiple second data rank sets with multiple first data rank sets.

[0051] Step S320: using a preset evaluation strategy to respectively determine the correlation between a plurality of first data ranked sets and a plurality of second data ranked sets having the same attribute.

[0052] Among them, the preset evaluation strategy can be to use the Pierman rank correlation coefficient to evaluate the monotonic relationship between the multiple first data ranking sets and the multiple second data ranking sets, and use the Kendall rank correlation coefficient to measure the rank correlation between the multiple first data ranking sets and the multiple second data ranking sets. Thus, the degree of correlation between the multiple first data ranking sets and the multiple second data ranking sets can be determined.

[0053] Step S330, determining a group of first data rank sets with the highest correlation with multiple second data rank sets as the target historical data set with the highest similarity to the second historical data of the target station.

[0054] Among them, after determining the degree of correlation between multiple first data ranked sets and multiple second data ranked sets, a group of first data ranked sets with the highest degree of correlation with multiple second data ranked sets are screened out from the multiple first data ranked sets, that is, the target historical data set with the highest similarity with the second historical data of the target station.

[0055] In this embodiment, through the rank set pair analysis method, the target historical data set with the highest similarity to the second historical data of the target station is screened out from the first historical data set. The target historical data set is equivalent to the data accumulated by the target station, thereby solving the problem of insufficient data accumulation of the target station.

[0056] In one embodiment, Figure 4As shown, in step S220, the target historical data set is used as a training sample to train the pre-built neural network model to determine the electricity price prediction model of the target site. Previously, the method for determining the electricity price prediction model also includes: steps S400-S410.

[0057] Step S400, constructing an initial neural network model.

[0058] Among them, the initial neural network model can be a deep autoregressive network (Deep Auto-Regression, DeepAR) model.

[0059] Step S410, using the first historical data set to train an initial neural network model to obtain a pre-built neural network model.

[0060] Among them, the first historical data set is the data set of the stations within the preset range of the target station. Since the stations within the preset range of the target station specifically refer to stations that have been established longer than the target station, and because the stations that were built and joined the market earlier have accumulated more data than the target station and have a larger amount of data accumulation, the data volume of the first historical data set is sufficient. The initial neural network model is trained using the first historical data set, and the pre-built neural network model obtained can accurately predict the electricity prices of the stations within the preset range of the target station.

[0061] The pre-built neural network model includes a likelihood function. After the neural network model is basically built, the likelihood function is added to make the output prediction value of the neural network model a probability distribution. Generally, Gaussian distribution (applicable to real continuous distribution data) or negative binomial distribution (applicable to integer data) can be selected.

[0062] Exemplarily, the process of training the initial neural network model using the first historical data set can be as follows: Figure 5 As shown, the input to each training unit is the data X corresponding to this moment in the first historical data set. i , t (independent variable) and the output of the entire model at the last moment Z i , t-1 The input will be passed to the hidden layer h at that moment i , t Perform training encoding. i , t Will be passed to the hidden layer h at the next moment i , t+1 For reference, the hidden layer update process is as follows:

[0063]

[0064] in, It refers to the given parameters of the Long Short-Term Memory (LSTM) cell state used in this autoregressive recurrent network. The LSTM unit converts the covariate x of the current time step t into i , t and the target variable z at the previous time step t-1 i , t-1 In addition, the LSTM receives the hidden state h of the previous time step i，t-1 The LSTM unit then outputs its hidden state h i , t , and send it to the next training step. The expression of the likelihood function can be:

[0065] In this embodiment, a training method for a pre-constructed neural network model is provided. The pre-constructed neural network model is trained using a first historical data set, so it can be regarded as a prediction model for stations within a preset range of the target station. Since the first historical data set contains a sufficient amount of data, the trained neural network model has a high prediction accuracy. In other words, a mature prediction model for stations within the preset range of the target station is obtained. This mature prediction model can be used as a pre-trained model for the target station in the future, which can save a lot of training time.

[0066] In one embodiment, Figure 6 As shown, step S220 uses the target historical data set as a training sample to train the pre-built neural network model to determine the electricity price prediction model of the target site. It includes: steps S600-S610.

[0067] Step S600, adding a fully connected layer on the basis of the pre-built neural network model to obtain a first electricity price prediction model.

[0068] Among them, a deep neural network (DNN) structure can be connected as a fully connected layer after the output of the pre-built neural network model.

[0069] Step S610, when the parameters of the pre-built neural network model remain unchanged, the first electricity price prediction model is trained using the target historical data set, and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site.

[0070] Among them, the parameters of the pre-built neural network model are first frozen, that is, in the process of training the first electricity price prediction model using the target historical data set, the parameters of the pre-built neural network model remain unchanged, thereby ensuring that the parameters of the pre-trained model remain unchanged. For example, the parameters of the pre-built neural network model can be stored as a variable table and set to remain unchanged. Taking the python implementation code as an example, the parameters of the model can be represented as mode l.parameters(), and the variables required_grad of all parameters in mode l.parameters() are set to False, which freezes the weights of these parameters. Only the parameters of the fully connected layer are trained, so that the fully connected layer can adapt to the electricity price prediction task, and the electricity price prediction model of the target site is obtained, which embodies the concept of transfer learning.

[0071] In this embodiment, by keeping the parameters of the pre-built neural network model fixed, the mature prediction model is utilized, and the parameters in the mature prediction model are kept unchanged, thereby keeping the trained model, and only the parameters of the fully connected layer are trained, so that the fully connected layer can adapt to the electricity price prediction task, and the electricity price prediction model of the target station is obtained, which embodies the concept of transfer learning, can save training resources, speed up the training speed, and thus obtain the electricity price prediction model of the target station more quickly.

[0072] In one embodiment, Figure 7 As shown, step S610, when the parameters of the pre-built neural network model are fixed, the first electricity price prediction model is trained using the target historical data set, and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site. It includes: steps S700-S710.

[0073] Step S700, when the parameters of the pre-built neural network model remain unchanged, the first electricity price prediction model is trained using the target historical data set, and the parameters of the fully connected layer are adjusted to obtain the second electricity price prediction model.

[0074] Among them, first, under the condition that the parameters of the pre-built neural network model are fixed, a small amount of data is used to train the first electricity price prediction model to obtain the second electricity price prediction model.

[0075] Step S710, using the target historical data set to train the second electricity price model, adjusting the parameters of the pre-built neural network model and the parameters of the fully connected layer, to obtain the electricity price prediction model of the target site.

[0076] Specifically, the parameters of the pre-built neural network model are unrestricted, but the hyperparameters such as the learning rate are adjusted down so that the pre-built neural network model has only a small change during the training process. The model is trained again and the accuracy of the model is verified to obtain the electricity price prediction model for the target site.

[0077] For example, the earlystop function can be used to stop training when the accuracy change is very small to prevent overfitting.

[0078] Among them, the adjustment range of the parameters of the pre-constructed neural network model is smaller than the preset range.

[0079] In this embodiment, while ensuring that the parameters of the pre-built neural network model are only fine-tuned, the electricity price prediction model of the target station is trained, which utilizes the concept of transfer learning, can save training resources, speed up the training speed, and thus obtain the electricity price prediction model of the target station more quickly.

[0080] In another embodiment of the present application, Figure 8 As shown, a method for predicting electricity price is provided, which is applied to the electricity price prediction model determined by the aforementioned method for determining the electricity price prediction model, and the method comprises: steps S800-S810.

[0081] Step S800, obtaining the data set of the target station for the day to be predicted.

[0082] Among them, the data set for the day to be predicted includes: meteorological data for the day to be predicted, power transmission data for the entire network for the day to be predicted, system load rate data for the day to be predicted, electricity price data for the entire network for the day to be predicted, installed capacity data for the entire network for the day to be predicted, and load data for the entire network for the day to be predicted.

[0083] Step S810, applying the electricity price prediction model, and determining the time series of the electricity price on the day to be predicted based on the data set of the day to be predicted.

[0084] In this embodiment, the electricity price prediction model is applied to realize the prediction of the time series of electricity price on the day to be predicted based on the data set of the day to be predicted, so as to determine the supply and demand relationship, market strategy, etc.

[0085] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0086] In this embodiment, a device for determining an electricity price prediction model is also provided, and the device for determining the electricity price prediction model is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0087] Fig. 9 is a structural block diagram of an optional device for determining an electricity price prediction model according to an embodiment of the present application. Fig. 9 As shown, including:

[0088] The data acquisition module 901 is used to acquire a first historical data set of at least one station within a preset range of the target station, wherein the first historical data set includes a time series of power data and environmental data in a one-to-one correspondence, and the power data includes at least power prices.

[0089] The rank construction module 902 is used to construct a rank set pair between the second historical data set and the first historical data set of the target station to determine the target historical data set with the highest similarity to the second historical data of the target station.

[0090] The training module 903 is used to use the target historical data set as a training sample to train the pre-built neural network model to determine the electricity price prediction model of the target site.

[0091] Through the above device, first obtain the first historical data set of at least one station within the preset range of the target station. Since the power generation conditions and the influencing factors of electricity price fluctuations of the new energy stations located closer are roughly the same, the historical data of the stations within the preset range of the target station can be used as a reference for the data of the target station. Then, a rank set pair is constructed between the second historical data set of the target station and the first historical data set, so that it is convenient to determine the target historical data set with the highest similarity to the second historical data of the target station through rank set pair analysis, and obtain the target historical data set that can best represent the historical data of the target station, which is equivalent to the data accumulation of the target station. Using the target historical data set as a training sample, the pre-constructed neural network model is trained to determine the electricity price prediction model of the target station, so that the data accumulated by the stations around the target station are analyzed to obtain the historical data that best meets the situation of the target station, and then the electricity price prediction model obtained by training the target historical data set can accurately predict the electricity price situation of the target station.

[0092] In an exemplary embodiment, the rank building module includes:

[0093] The first data rank set determination unit is used to sort the elements of each attribute in the first historical data set according to a preset rank to obtain multiple first data rank sets of the first historical data set, wherein different first data rank sets contain elements with different attributes.

[0094] The second data rank set determination unit is used to sort the elements of each attribute in the second historical data set according to a preset rank to obtain multiple second data rank sets of the second historical data set.

[0095] The correlation degree determination unit is used to adopt a preset evaluation strategy to respectively determine the correlation degrees between multiple first data ranked sets and multiple second data ranked sets with the same attributes.

[0096] The target historical data set determination unit is used to determine a group of first data ranking sets with the highest correlation with multiple second data ranking sets as the target historical data set with the highest similarity with the second historical data of the target station.

[0097] In an exemplary embodiment, the above device further includes:

[0098] The initial neural network model determination unit is used to construct the initial neural network model.

[0099] The first training unit is used to train an initial neural network model using a first historical data set to obtain a pre-constructed neural network model, wherein the pre-constructed neural network model includes a likelihood function.

[0100] In an exemplary embodiment, the first training unit includes:

[0101] The second training unit is used to add a fully connected layer based on the pre-built neural network model to obtain a first electricity price prediction model.

[0102] The third training unit is used to train the first electricity price prediction model using the target historical data set while keeping the parameters of the pre-built neural network model fixed, adjust the parameters of the fully connected layer, and obtain the electricity price prediction model of the target site.

[0103] In an exemplary embodiment, the third training unit includes:

[0104] The fourth training unit is used to train the first electricity price prediction model using the target historical data set while keeping the parameters of the pre-built neural network model fixed, adjust the parameters of the fully connected layer, and obtain the second electricity price prediction model.

[0105] The fifth training unit is used to train the second electricity price model using the target historical data set, adjust the parameters of the pre-constructed neural network model and the parameters of the fully connected layer, and obtain the electricity price prediction model of the target site, wherein the adjustment range of the parameters of the pre-constructed neural network model is smaller than the preset range.

[0106] An embodiment of the present application further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0107] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0108] S1, obtaining a first historical data set of at least one station within a preset range of a target station, wherein the first historical data set includes a time series of power data and environmental data in one-to-one correspondence, and the power data includes at least power prices.

[0109] S2, constructing a rank set pair between the second historical data set and the first historical data set of the target station to determine a target historical data set having the highest similarity to the second historical data of the target station.

[0110] S3, using the target historical data set as a training sample, trains the pre-built neural network model to determine the electricity price prediction model of the target site.

[0111] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0112] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0113] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0114] S1, obtaining a first historical data set of at least one station within a preset range of a target station, wherein the first historical data set includes a time series of power data and environmental data in one-to-one correspondence, and the power data includes at least power prices.

[0115] S2, constructing a rank set pair between the second historical data set and the first historical data set of the target station to determine a target historical data set having the highest similarity to the second historical data of the target station.

[0116] S3, using the target historical data set as a training sample, trains the pre-built neural network model to determine the electricity price prediction model of the target site.

[0117] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0118] An embodiment of the present application further provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores the computer program product, and when the computer program is executed by a processor, the steps of the method in each embodiment of the present application are implemented.

[0119] Optionally, in this embodiment, the above computer program may be configured to implement the following steps when executed by a processor:

[0120] S1, obtaining a first historical data set of at least one station within a preset range of a target station, wherein the first historical data set includes a time series of power data and environmental data in one-to-one correspondence, and the power data includes at least power prices.

[0121] S2, constructing a rank set pair between the second historical data set and the first historical data set of the target station to determine a target historical data set having the highest similarity to the second historical data of the target station.

[0122] S3, using the target historical data set as a training sample, trains the pre-built neural network model to determine the electricity price prediction model of the target site.

[0123] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0124] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0125] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining an electricity price prediction model, characterized in that: The method comprises: Acquire a first historical data set of at least one station within a preset range of the target station, wherein the first historical data set includes a time series of power data and environmental data corresponding to each other, and the power data includes at least power price; Constructing a rank set pair between the second historical data set of the target station and the first historical data set to determine a target historical data set with the highest similarity to the second historical data of the target station; The target historical data set is used as a training sample to train a pre-built neural network model to determine an electricity price prediction model for the target site.

2. The method according to claim 1, characterized in that The step of constructing a rank set pair between the second historical data set of the target station and the first historical data set to determine a target historical data set having the highest similarity to the second historical data of the target station includes: Sort the elements of each attribute in the first historical data set according to a preset rank, to obtain a plurality of first data rank sets of the first historical data set, wherein different first data rank sets contain elements with different attributes; Sort the elements of each attribute in the second historical data set according to the preset rank, to obtain a plurality of second data rank sets of the second historical data set; Using a preset evaluation strategy, respectively determining the correlation between the plurality of first data ranked sets and the plurality of second data ranked sets having the same attribute; A group of first data ranking sets with the highest correlation with the multiple second data ranking sets is determined as the target historical data set with the highest similarity with the second historical data of the target site.

3. The method according to claim 1 or 2, characterized in that: Before using the target historical data set as a training sample to train a pre-built neural network model to determine the electricity price prediction model of the target site, the method further includes: Build an initial neural network model; The initial neural network model is trained using the first historical data set to obtain the pre-constructed neural network model, wherein the pre-constructed neural network model includes a likelihood function.

4. The method according to claim 3, characterized in that The method of using the target historical data set as a training sample to train a pre-built neural network model to determine the electricity price prediction model of the target site includes: Adding a fully connected layer on the basis of the pre-built neural network model to obtain a first electricity price prediction model; When the parameters of the pre-built neural network model remain fixed, the first electricity price prediction model is trained using the target historical data set, and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site.

5. The method according to claim 4, characterized in that In the case where the parameters of the pre-built neural network model remain unchanged, the first electricity price prediction model is trained using the target historical data set, and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site, including: In the case where the parameters of the pre-built neural network model remain unchanged, the first electricity price prediction model is trained using the target historical data set, and the parameters of the fully connected layer are adjusted to obtain a second electricity price prediction model; The target historical data set is used to train the second electricity price model, and the parameters of the pre-constructed neural network model and the parameters of the fully connected layer are adjusted to obtain the electricity price prediction model of the target site, wherein the adjustment range of the parameters of the pre-constructed neural network model is smaller than the preset range.

6. A method for predicting electricity prices, characterized in that: An electricity price prediction model determined by the method for determining an electricity price prediction model according to any one of claims 1 to 5, the method comprising: Obtain the data set of the target station for the day to be predicted; The electricity price prediction model is applied to determine the time series of the electricity price on the day to be predicted based on the data set of the day to be predicted.

7. A device for determining an electricity price prediction model, characterized in that: The device comprises: A data acquisition module, configured to acquire a first historical data set of at least one station within a preset range of the target station, wherein the first historical data set includes a time series of power data and environmental data corresponding to each other, and the power data includes at least power price; A rank construction module, used to construct a rank set pair between the second historical data set of the target station and the first historical data set, so as to determine a target historical data set with the highest similarity to the second historical data of the target station; The training module is used to use the target historical data set as a training sample to train a pre-built neural network model to determine the electricity price prediction model of the target site.

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

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images