Settlement parameter prediction method and device, storage medium and electronic equipment

Abstract

The invention discloses a settlement parameter prediction method and device, a storage medium and electronic equipment. The method comprises the steps that period attribute data of a to-be-predicted target settlement period is acquired, and the period duration of the target settlement period is larger than or equal to a target duration threshold value; in a settlement parameter prediction network, target settlement parameters corresponding to the target settlement period are predicted based on settlement time sequence characteristics extracted from the period attribute data, and the settlement parameter prediction network comprises a double-layer convolution sub-network and a double-layer feedforward sub-network; the settlement parameter prediction network is a neural network which uses the settlement data sample to carry out iteration training for multiple times until a convergence condition is reached, and the convergence condition indicates that a training loss value output by the settlement parameter prediction network in training reaches a set loss condition value. The technical problem that the accuracy of a prediction result is difficult to ensure due to the fact that short-term and time-efficient disclosed data are difficult to capture medium-and-long-term parameter change trends is solved.

Description

Technical Field

[0001] The present application relates to the field of power systems, and in particular to a method and device for predicting settlement parameters, a storage medium, and an electronic device. Background Art

[0002] In order to count and control the electricity consumption of users in a given area, the power system adopts a prediction mechanism for power settlement parameters. However, the power settlement parameters in actual operation are highly volatile, especially the prediction process for medium and long periods (such as 7 to 14 days) faces major challenges.

[0003] Existing forecasting methods are highly dependent on scheduling disclosure data, and the data before scheduling disclosure is only disclosed on D+1 day on D day, which has only short-term timeliness and can only assist in predicting settlement parameters for 1 to 2 days, and it is difficult to accurately capture the trend of medium and long cycles. In addition, if this short-term disclosure data is used to deal with complex time series patterns and the forecast process of multiple influencing factors at the same time, it is difficult to ensure the accuracy of the forecast results of settlement parameters due to the limitation of timeliness.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present application provide a method and device for predicting settlement parameters, a storage medium and an electronic device, so as to at least solve the technical problem that it is difficult to ensure the accuracy of the prediction results due to the difficulty of short-term disclosure data in capturing medium- and long-term parameter change trends.

[0006] According to one aspect of an embodiment of the present application, a method for predicting settlement parameters is provided, comprising: obtaining period attribute data of a target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to a target duration threshold; predicting a target settlement parameter corresponding to the target settlement period based on settlement timing features extracted from the period attribute data in a settlement parameter prediction network, wherein the settlement parameter prediction network comprises a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained for multiple times using settlement data samples until a convergence condition is reached, and the convergence condition indicates that a training loss value output by the settlement parameter prediction network during training has reached a set loss condition value.

[0007] Optionally, in this embodiment, the above-mentioned acquisition of the period attribute data of the target settlement period to be predicted includes: acquiring the meteorological data of the above-mentioned target settlement period, and the electricity consumption disclosure conditions configured for the electric equipment during the above-mentioned target settlement period, wherein the above-mentioned meteorological data is the data collected during the above-mentioned target settlement period to reflect the changes in natural weather, and the above-mentioned electricity consumption disclosure conditions include: the load of the above-mentioned electric equipment, the capacity information of the above-mentioned electric equipment, and the electricity consumption strategy of the above-mentioned electric equipment; based on the above-mentioned meteorological data and the above-mentioned electricity consumption disclosure conditions, the above-mentioned period attribute data matching the above-mentioned target settlement period is determined.

[0008] Optionally, in this embodiment, in the settlement parameter prediction network, based on the settlement timing characteristics extracted from the above-mentioned period attribute data, predicting the target settlement parameters corresponding to the above-mentioned target settlement period includes: extracting the above-mentioned settlement timing characteristics corresponding to the above-mentioned target settlement period from the above-mentioned period attribute data in the above-mentioned settlement parameter prediction network; predicting the above-mentioned target settlement parameters of the above-mentioned target settlement period based on the above-mentioned settlement timing characteristics and historical settlement timing characteristics corresponding to the historical settlement periods in the above-mentioned settlement parameter prediction network.

[0009] Optionally, in this embodiment, before obtaining the period attribute data of the target settlement period to be predicted, it also includes: inputting the above-mentioned settlement data sample into the initialized above-mentioned settlement parameter prediction network for training to obtain a training result; inputting the current label settlement parameter corresponding to the above-mentioned settlement data sample currently input and the above-mentioned training result currently output into the loss function to obtain a current training loss value; when the above-mentioned current training loss value has not yet reached the above-mentioned set loss condition value, adjusting the network parameters in the above-mentioned settlement parameter prediction network in training; when the above-mentioned current training loss value reaches the above-mentioned set loss condition value, determining that the above-mentioned settlement parameter prediction network in training has reached the above-mentioned convergence condition.

[0010] Optionally, in this embodiment, before the above-mentioned acquisition of the period attribute data of the target settlement period to be predicted, it also includes: acquiring historical settlement data obtained within the historical settlement period, historical electricity consumption disclosure conditions configured for the power-consuming equipment within the above-mentioned historical settlement period, and reference meteorological data of a reference settlement period associated with the above-mentioned historical settlement period, wherein the above-mentioned reference settlement period is a predicted settlement period associated with the corresponding above-mentioned historical settlement period; normalizing the above-mentioned historical settlement data, the above-mentioned historical electricity consumption disclosure conditions and the above-mentioned reference meteorological data to obtain multiple input data sets, and using the above-mentioned multiple input data sets as the above-mentioned settlement data samples.

[0011] Optionally, in this embodiment, before obtaining the period attribute data of the target settlement period to be predicted, it also includes: associating and connecting the data segmentation subnetwork, the above-mentioned two-layer convolution subnetwork and the above-mentioned two-layer feedforward subnetwork to construct the above-mentioned settlement parameter prediction network, wherein the above-mentioned data segmentation subnetwork is used to perform block processing on the input settlement data samples, the above-mentioned two-layer convolution subnetwork includes: a deep convolution subnetwork and a point convolution subnetwork, and the above-mentioned two-layer feedforward subnetwork is used for regression processing.

[0012] Optionally, in this embodiment, the above-mentioned settlement data samples are input into the initialized settlement parameter prediction network for training, and the training results obtained include: the above-mentioned settlement data samples currently input are divided into blocks in the above-mentioned data segmentation subnetwork to obtain multiple sample data blocks that meet the dividing conditions; the current sample settlement timing features of the above-mentioned multiple sample data blocks are obtained through the above-mentioned double-layer convolution subnetwork, wherein the above-mentioned current sample settlement timing features include: the intra-block timing relationship features determined by the above-mentioned deep convolution subnetwork, and the inter-block timing relationship features determined by the above-mentioned point convolution subnetwork; the above-mentioned current sample settlement timing features are regressed through the above-mentioned double-layer feedforward subnetwork to obtain the above-mentioned training result currently output.

[0013] According to another aspect of an embodiment of the present application, a settlement parameter prediction device is also provided, including: a first acquisition unit, used to acquire period attribute data of a target settlement period to be predicted, wherein the period duration of the above-mentioned target settlement period is greater than or equal to a target duration threshold; a prediction unit, used to predict the target settlement parameters corresponding to the above-mentioned target settlement period based on the settlement timing characteristics extracted from the above-mentioned period attribute data in a settlement parameter prediction network, wherein the above-mentioned settlement parameter prediction network includes a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the above-mentioned settlement parameter prediction network is a neural network that is iteratively trained for multiple times using settlement data samples until a convergence condition is reached, and the above-mentioned convergence condition indicates that the training loss value output by the above-mentioned settlement parameter prediction network during training has reached a set loss condition value.

[0014] According to another aspect of the embodiments of the present application, a storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned settlement parameter prediction method when running.

[0015] According to another aspect of the embodiments of the present application, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the above settlement parameter prediction method.

[0016] 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 settlement parameter prediction method through the computer program.

[0017] In an embodiment of the present application, a settlement parameter prediction network including a double-layer convolutional subnetwork and a double-layer feedforward subnetwork is used to perform predictive analysis on the periodic attribute data of the target settlement period to be predicted, so as to obtain the target settlement parameters that match the target settlement period, thereby achieving the prediction of the corresponding medium- and long-term settlement parameters by learning the medium- and long-term characteristics of the settlement period itself, so as to cope with the complex changes faced by energy in the medium and long term, improve the accuracy of settlement predictions, and thus solve the technical problem that it is difficult to ensure the accuracy of the prediction results due to the difficulty of short-term disclosure data in capturing the medium- and long-term parameter change trends. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 is a flow chart of an optional settlement parameter prediction method according to an embodiment of the present application;

[0020] Figure 2 is a flowchart of another optional settlement parameter prediction method according to an embodiment of the present application;

[0021] Figure 3 is a schematic diagram of an optional settlement parameter prediction method according to an embodiment of the present application;

[0022] Figure 4 is a schematic structural diagram of an optional settlement parameter prediction device according to an embodiment of the present application;

[0023] Figure 5 It is a schematic diagram of the structure of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the present application, the technical solutions 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 this application.

[0025] 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 where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device 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 devices.

[0026] According to one aspect of the embodiments of the present application, a method for predicting a settlement parameter is provided, such as Figure 1 As shown, the method includes:

[0027] S102, obtaining period attribute data of a target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to a target duration threshold;

[0028] Optionally, in this embodiment, the target settlement period may be, but is not limited to, a statistical period for delivering an equivalent price to the entity providing natural energy for the natural energy used. For example, the natural energy here may include, but is not limited to: electric energy, water energy, gas energy, etc.

[0029] In addition, the above-mentioned periodic attribute data may include, but are not limited to: meteorological data within the target settlement period, and electricity consumption disclosure conditions corresponding to the target settlement period, wherein the above-mentioned meteorological data may include data collected within the target settlement period to reflect natural weather changes, such as temperature, irradiance, precipitation, and wind speed. The above-mentioned electricity consumption disclosure conditions may include, but are not limited to: load of electrical equipment, capacity information of electrical equipment, and electricity consumption strategies of electrical equipment, for example, electricity consumption strategies include: new energy output, west-to-east delivery, etc.

[0030] S104, in the settlement parameter prediction network, based on the settlement timing features extracted from the period attribute data, predict the target settlement parameters corresponding to the target settlement period, wherein the settlement parameter prediction network includes a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained multiple times using settlement data samples until a convergence condition is reached, and the convergence condition indicates that the training loss value output by the settlement parameter prediction network in training has reached a set loss condition value.

[0031] Optionally, in this embodiment, the network structure of the settlement parameter prediction network may include, but is not limited to: a data segmentation subnetwork, a two-layer convolution subnetwork and a two-layer feedforward subnetwork. Among them, the data segmentation subnetwork is used to uniformly block the input data according to a set format; the two-layer convolution subnetwork includes two layers of convolution subnetworks, both of which are used to extract features and learn the block data obtained after the block division to obtain time series features; the two-layer feedforward subnetwork may be used, but is not limited to, for regression processing of the results output by the two-layer convolution subnetwork to obtain prediction results. It should be noted that the two-layer convolution subnetwork is a two-layer separable convolution subnetwork. Separable convolution is a convolution operation, which is divided into two stages: depth convolution and point-by-point convolution. It can effectively reduce the number of parameters and the amount of calculation, and improve the performance and generalization ability of the network. In the two-layer separable convolution subnetwork, the two separable convolution layers are usually connected together through operations such as activation functions and pooling layers to extract higher-level feature representations.

[0032] In addition, in this embodiment, the above-mentioned settlement data samples can be but are not limited to the input data set obtained after preprocessing the original training data. For example, taking electric energy as an example, the input data set here may include the following types: electricity prices, past covariates (without future medium- and long-term forecast data) and future covariates (with future medium- and long-term forecasts or statistical data), among which the disclosed boundary conditions are past covariates and meteorological data are future covariates.

[0033] It should be noted that, using the settlement data samples, in the process of training in the above-built network architecture, it is possible but not limited to obtain the loss value of the training data, and optimize the network parameters (or network weights) in the above-mentioned settlement parameter prediction network through the back propagation algorithm based on the loss value. This method is used to evaluate the prediction performance of the above-mentioned settlement parameter prediction network, and adjust and optimize the model parameters according to the situation to ensure that the model has good prediction capabilities. Still taking electricity as an example, after the settlement parameter prediction network is trained, the settlement parameter prediction network can be used to output the long-period sequence in the day-ahead electricity price, so as to accurately predict the electricity price trend over a longer time range.

[0034] For example, specific combination Figure 2 The example shown is used to illustrate:

[0035] The cycle attribute data 201 of the target settlement cycle to be predicted is obtained, and then the cycle attribute data 201 is input into the settlement parameter prediction network 202, wherein the settlement parameter prediction network 202 includes: a data segmentation subnetwork 2021, a two-layer convolution subnetwork 2022, and a two-layer feedforward subnetwork 2023. After a series of processing such as block processing, feature extraction and learning are performed on the cycle attribute data 201, the corresponding settlement timing features are obtained, and then the target settlement parameters 203 are output based on the cycle timing features.

[0036] Through the embodiments provided by the present application, a settlement parameter prediction network including a double-layer convolutional subnetwork and a double-layer feedforward subnetwork is used to perform predictive analysis on the periodic attribute data of the target settlement period to be predicted, so as to obtain the target settlement parameters that match the target settlement period, thereby achieving the prediction of the corresponding medium- and long-term settlement parameters by learning the medium- and long-term characteristics of the settlement period itself, so as to cope with the complex changes faced by energy in the medium and long term, improve the accuracy of settlement predictions, and thus overcome the technical problem of low accuracy of prediction results caused by the short-term timeliness of disclosed data in related technologies.

[0037] As an optional solution, obtaining the period attribute data of the target settlement period to be predicted includes:

[0038] S1, obtaining meteorological data of the target settlement period and electricity consumption disclosure conditions configured for the electricity-consuming equipment during the target settlement period, wherein the meteorological data is data collected during the target settlement period to reflect natural weather changes, and the electricity consumption disclosure conditions include: load of the electricity-consuming equipment, capacity information of the electricity-consuming equipment, and electricity consumption strategy of the electricity-consuming equipment;

[0039] S2, based on meteorological data and electricity consumption disclosure conditions, determines the period attribute data that matches the target settlement period.

[0040] Optionally, in this embodiment, the above-mentioned periodic attribute data may include but are not limited to: meteorological data and electricity consumption disclosure conditions, wherein the meteorological data may be but are not limited to future covariates, that is, assuming that the target settlement period to be predicted is D-day, then the meteorological data may be meteorological data from D+1 to D+14 (such as temperature, irradiance, precipitation, wind speed, etc.). The electricity consumption disclosure conditions may be but are not limited to past covariates, that is, assuming that the target settlement period to be predicted is D-day, then the electricity consumption disclosure conditions may be disclosure boundary conditions from D-day to D-1 (such as load, electricity consumption strategy, capacity information, etc.).

[0041] In addition, in this embodiment, after obtaining the meteorological data and electricity consumption disclosure conditions, the data can also be normalized but not limited to obtain an input data set in a standard format to facilitate input of the data in a standard format into the settlement parameter prediction network.

[0042] For example, still taking electric energy data as an example, after obtaining the electric energy cycle attribute data, it is divided into blocks according to the preset block length D, and the block step is S. After obtaining multiple block data, they are stacked to obtain a 1*D*N matrix, where N is the number of settlement cycles.

[0043] Through the embodiments provided in the present application, in order to predict the target settlement parameters corresponding to the target settlement period, the meteorological data and electricity consumption disclosure conditions of the target settlement period are used in advance to construct the periodic attribute data of the target settlement period, thereby facilitating the settlement parameter prediction network to directly predict the settlement parameters of the target settlement period, reducing the data preprocessing links deployed in the settlement parameter prediction network, simplifying the network architecture of the settlement parameter prediction network, and thereby improving the processing efficiency of the settlement parameter prediction network.

[0044] As an optional solution, in the settlement parameter prediction network, based on the settlement time series features extracted from the period attribute data, the target settlement parameters corresponding to the target settlement period are predicted to include:

[0045] S1, extracting the settlement time series features corresponding to the target settlement period from the period attribute data in the settlement parameter prediction network;

[0046] S2, predicting the target settlement parameters of the target settlement period in the settlement parameter prediction network based on the settlement timing characteristics and the historical settlement timing characteristics corresponding to the historical settlement period.

[0047] Optionally, in this embodiment, in the settlement parameter prediction network, the above-mentioned periodic attribute data can be processed in blocks but is not limited to, and combined and stacked to obtain a data matrix, such as a 1*D*N data block. Then the above-mentioned block data (also referred to as data blocks) is processed by a linear head and a nonlinear head, wherein the linear head is a residual connection. The nonlinear head includes a two-layer separable convolutional neural network: the first layer is a depth convolution, in order to capture the timing relationship within the block; the second layer is a point convolution, in order to capture the timing relationship between the blocks; the two layers of convolution are also connected by residuals. Regression processing is performed on the settlement timing characteristics predicted and output by the above-mentioned two-layer separable convolutional neural network to obtain the final target settlement parameters.

[0048] It should be noted that the network parameters in the above settlement parameter prediction network are determined after multiple iterations of training based on the historical settlement time series characteristics corresponding to the historical settlement cycle. That is to say, in the settlement parameter prediction network, the settlement time series characteristics corresponding to the currently input target settlement cycle are predicted using the network parameters (network weights) obtained through training, and the settlement parameters corresponding to the target settlement cycle will be obtained. This will achieve accurate prediction of the medium and long-term settlement trend.

[0049] Through the embodiments provided by the present application, in a settlement parameter prediction network that determines network parameters by utilizing historical settlement timing characteristics corresponding to historical settlement cycles, target settlement parameters are predicted for the target settlement cycle based on settlement timing characteristics corresponding to the target settlement cycle extracted from period attribute data, thereby facilitating accurate analysis of medium- and long-term energy usage trends and energy settlement trends through timing characteristics, thereby assisting the power consumption control end in making more accurate energy allocation decisions and avoiding energy waste.

[0050] As an optional solution, before obtaining the period attribute data of the target settlement period to be predicted, the following is also included:

[0051] S1, input the settlement data sample into the initialized settlement parameter prediction network for training to obtain the training result;

[0052] S2, input the current label settlement parameter corresponding to the currently input settlement data sample and the currently output training result into the loss function to obtain the current training loss value;

[0053] S3-1, when the current training loss value has not reached the set loss condition value, adjusting the network parameters in the settlement parameter prediction network during training;

[0054] S3-2, when the current training loss value reaches the set loss condition value, it is determined that the settlement parameter prediction network in training has reached the convergence condition.

[0055] As an optional implementation, before obtaining the period attribute data of the target settlement period to be predicted, it also includes: obtaining historical settlement data obtained within the historical settlement period, historical electricity consumption disclosure conditions configured for the power-consuming equipment within the historical settlement period, and reference meteorological data of a reference settlement period associated with the historical settlement period, wherein the reference settlement period is a predicted settlement period associated with the corresponding historical settlement period; normalizing the historical settlement data, historical electricity consumption disclosure conditions and reference meteorological data to obtain multiple input data sets, and using the multiple input data sets as settlement data samples.

[0056] For example, let's still take electric energy data as an example. Furthermore, let's assume that the historical settlement data obtained in the historical settlement period is the settlement data generated in the past 90 days, such as the historical electricity price, that is, the electricity price from D-90 to D. Let's assume that the historical electricity disclosure conditions configured for the power-consuming equipment in the historical settlement period are load, capacity information, and electricity strategy (such as new energy output, west-to-east transmission, etc.). Let's assume that the reference meteorological data of the reference settlement period associated with the historical settlement period is the meteorological data from D+1 to D+14 (temperature, irradiance, precipitation, wind speed).

[0057] The above input data are classified to obtain the following three types of input data sets: electricity prices, past covariates (without future medium- and long-term forecast data), and future covariates (with future medium- and long-term forecasts or statistical data), among which the disclosed boundary conditions are past covariates and meteorological data are future covariates. Then the above input data sets are normalized to obtain settlement data samples in a standardized format.

[0058] As an optional implementation, before obtaining the periodic attribute data of the target settlement period to be predicted, it also includes: associating and connecting the data segmentation subnetwork, the two-layer convolution subnetwork and the two-layer feedforward subnetwork to construct a settlement parameter prediction network, wherein the data segmentation subnetwork is used to perform block processing on the input settlement data samples, the two-layer convolution subnetwork includes: a deep convolution subnetwork and a point convolution subnetwork, and the two-layer feedforward subnetwork is used for regression processing.

[0059] Optionally, in this embodiment, the above two-layer convolution subnetwork is a two-layer separable convolution subnetwork, which includes: the first layer is a deep convolution subnetwork for capturing the temporal relationship within the block; the second layer is a point convolution subnetwork for capturing the temporal relationship between blocks. The two layers of convolution are also connected by residuals.

[0060] Among them, the deep convolution applies convolution to each input channel independently to capture the internal temporal feature relationship of the block. The formula is:

[0061] x (l) =BN(σ(Conv(x (l-1) ,k=K))) (1)

[0062] Among them, the point convolution performs 1x1 convolution on each position, which is mainly used to integrate the information between channels and capture the temporal relationship between blocks. The formula is:

[0063] x (l) =BN(σ(Conv(x (l-1) ,k=1))) (2)

[0064] Among them, BN means batch normalization;

[0065] σ represents the activation function, using GELU, the GELU formula is:

[0066]

[0067] Conv represents the convolution operation, k is the convolution kernel size;

[0068] x (l-1) is the convolution input data block, x (l) is the convolution output data block.

[0069] As an optional implementation, S1, inputting the settlement data sample into the initialized settlement parameter prediction network for training, and obtaining the training results including:

[0070] S11, performing block processing on the currently input settlement data sample in the data segmentation sub-network to obtain multiple sample data blocks that meet the block conditions;

[0071] S12, obtaining current sample settlement timing features of multiple sample data blocks through a double-layer convolution subnetwork, wherein the current sample settlement timing features include: intra-block timing relationship features determined by the deep convolution subnetwork, and inter-block timing relationship features determined by the point convolution subnetwork;

[0072] S13, regressing the current sample settlement time series features through a double-layer feedforward sub-network to obtain the current output training result.

[0073] It should be noted that the above-mentioned block conditions may be, but are not limited to, block size and block step size. For example, still taking electric energy data as an example, the above-mentioned input data set for block processing includes: electricity price and past covariates.

[0074] For example, let's still take electric energy data as an example. Further, let's assume that the historical settlement data obtained in the historical settlement period is the settlement data generated in the past 90 days, such as the historical electricity price, that is, the electricity price from D-90 to D. Let's assume that the historical electricity disclosure conditions configured for the power-consuming equipment in the historical settlement period are load, capacity information, and electricity strategy (such as new energy output, west-to-east transmission, etc.). Let's assume that the reference meteorological data of the reference settlement period associated with the historical settlement period is the meteorological data from D+1 to D+14 (temperature, irradiance, precipitation, wind speed). The input data is classified to obtain the following three types of input data sets: electricity price, past covariates (no future medium- and long-term forecast data), and future covariates (with future medium- and long-term forecasts or statistical data), where the disclosed boundary conditions are past covariates and the meteorological data are future covariates.

[0075] Then, if Figure 3 As shown in the figure, the electricity price and past covariate data in the above input data set are divided into multiple patches, and the length of a single day in the time series (96 electricity prices) is taken as the length of the patch D, and the patch step S is also 96 (no overlap in the patches). For these data, the size of the stacked matrices after patching is 1*D*N, where N is 90, which is the number of days of training data.

[0076] Furthermore, the divided multiple sample data blocks are input into the above-mentioned two-layer separable convolution subnetwork. As shown in the figure, the deep convolution subnetwork and the point convolution subnetwork are respectively input, and then the two-layer convolution processing results are residually linked to obtain the forecast output from D+1 to D+14. In addition, the future covariate data from D+1 to D+14 (such as the reference meteorological data of the reference settlement period associated with the above-mentioned historical settlement period) are added and uniformly input into the two-layer feedforward neural network for regression training, in which the activation function is GELU. Finally, the output of the day-ahead electricity price from D+1 to D+14 is obtained (that is, the training result of the current output).

[0077] Training is performed in the settlement parameter prediction network of the above network architecture, and the training loss value output by the corresponding loss function is obtained. Then, the network parameters (i.e., network weights) in the settlement parameter prediction network are optimized through the back propagation optimization algorithm based on the training loss value.

[0078] Through the embodiments provided in this application, data is processed using block, double-layer separable convolution, and feedforward neural networks to predict future medium- and long-term settlement parameters. This effectively predicts medium- and long-term settlement parameters, facilitates understanding of the changing trends of medium- and long-term settlement parameters, and facilitates market participants to more accurately provide medium- and long-term energy allocation decisions, improve energy utilization, and avoid unnecessary energy waste.

[0079] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0080] According to another aspect of the embodiments of the present application, a settlement parameter prediction device for implementing the above settlement parameter prediction method is also provided. Figure 4 As shown, the device comprises:

[0081] The first acquisition unit 402 is used to acquire the period attribute data of the target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to the target duration threshold;

[0082] The prediction unit 404 is used to predict the target settlement parameters corresponding to the target settlement period based on the settlement timing characteristics extracted from the period attribute data in the settlement parameter prediction network, wherein the settlement parameter prediction network includes a two-layer convolution subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained multiple times using settlement data samples until a convergence condition is reached, and the convergence condition indicates that the training loss value output by the settlement parameter prediction network in training has reached a set loss condition value.

[0083] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0084] As an optional solution, the first obtaining unit 402 includes:

[0085] An acquisition module is used to acquire meteorological data of a target settlement period and electricity consumption disclosure conditions configured for electric equipment during the target settlement period, wherein meteorological data is data collected during the target settlement period to reflect natural weather changes, and electricity consumption disclosure conditions include: load of electric equipment, capacity information of electric equipment, and electricity consumption strategy of electric equipment;

[0086] The determination module is used to determine the cycle attribute data matching the target settlement cycle based on meteorological data and electricity consumption disclosure conditions.

[0087] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0088] As an optional solution, the prediction unit 404 includes:

[0089] An extraction module, used to extract settlement timing characteristics corresponding to a target settlement period from period attribute data in a settlement parameter prediction network;

[0090] The prediction module is used to predict the target settlement parameters of the target settlement period based on the settlement timing characteristics and the historical settlement timing characteristics corresponding to the historical settlement period in the settlement parameter prediction network.

[0091] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0092] As an optional solution, it also includes:

[0093] A training unit, used for inputting settlement data samples into the initialized settlement parameter prediction network for training before obtaining the period attribute data of the target settlement period to be predicted, and obtaining a training result;

[0094] A processing unit, used to input the current label settlement parameter corresponding to the currently input settlement data sample and the currently output training result into the loss function to obtain the current training loss value;

[0095] An adjustment unit, used to adjust the network parameters in the settlement parameter prediction network in training when the current training loss value has not reached the set loss condition value;

[0096] The determination unit is used to determine that the settlement parameter prediction network in training has reached the convergence condition when the current training loss value reaches the set loss condition value.

[0097] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0098] As an optional solution, it also includes:

[0099] The second acquisition unit is used to acquire the historical settlement data obtained in the historical settlement period, the historical electricity consumption disclosure conditions configured for the electric equipment in the historical settlement period, and the reference meteorological data of the reference settlement period associated with the historical settlement period before acquiring the period attribute data of the target settlement period to be predicted, wherein the reference settlement period is the predicted settlement period associated with the corresponding historical settlement period;

[0100] The normalization processing unit is used to normalize the historical settlement data, the historical electricity consumption disclosure conditions and the reference meteorological data to obtain multiple input data sets, and use the multiple input data sets as settlement data samples.

[0101] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0102] As an optional solution, it also includes:

[0103] A construction unit is used to associate and connect the data segmentation subnetwork, the double-layer convolution subnetwork and the double-layer feedforward subnetwork before obtaining the period attribute data of the target settlement period to be predicted, so as to construct a settlement parameter prediction network, wherein the data segmentation subnetwork is used to perform block processing on the input settlement data samples, the double-layer convolution subnetwork includes: a deep convolution subnetwork and a point convolution subnetwork, and the double-layer feedforward subnetwork is used for regression processing.

[0104] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0105] As an optional option, the training unit includes:

[0106] A block division module is used to perform block processing on the currently input settlement data sample in the data segmentation sub-network to obtain multiple sample data blocks that meet the block division conditions;

[0107] A convolution calculation module, used to obtain the current sample settlement timing features of multiple sample data blocks through a double-layer convolution subnetwork, wherein the current sample settlement timing features include: intra-block timing relationship features determined by the deep convolution subnetwork and inter-block timing relationship features determined by the point convolution subnetwork;

[0108] The regression processing module is used to perform regression processing on the current sample settlement time series features through a double-layer feedforward sub-network to obtain the current output training result.

[0109] Optionally, the solution in this embodiment may be but is not limited to referring to the above method embodiment, which will not be described in detail here.

[0110] According to another aspect of the embodiment of the present application, an electronic device for implementing the above settlement parameter prediction method is also provided. Figure 5 As shown, the electronic device includes a memory 502 and a processor 504. The memory 502 stores a computer program, and the processor 504 is configured to execute the steps in any of the above method embodiments through the computer program.

[0111] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0112] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0113] S1, obtaining period attribute data of a target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to a target duration threshold;

[0114] S2, in the settlement parameter prediction network, based on the settlement timing features extracted from the period attribute data, predict the target settlement parameters corresponding to the target settlement period, wherein the settlement parameter prediction network includes a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained multiple times using settlement data samples until the convergence condition is reached, and the convergence condition indicates that the training loss value output by the settlement parameter prediction network in training has reached the set loss condition value.

[0115] Alternatively, a person skilled in the art may understand that: Figure 5The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 5 The structure of the electronic device is not limited. Figure 5 More or fewer components (such as network interfaces, etc.) as shown in, or with Figure 5 Different configurations shown.

[0116] Among them, the memory 502 can be used to store software programs and modules, such as the program instructions / modules corresponding to the settlement parameter prediction method and device in the embodiment of the present application. The processor 504 executes various functional applications and data processing by running the software programs and modules stored in the memory 502, that is, realizing the above-mentioned settlement parameter prediction method. The memory 502 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 instances, the memory 502 may further include a memory remotely located relative to the processor 504, and these remote memories may be connected to the 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. As an example, Figure 5 As shown, the memory 502 may include but is not limited to the first acquisition unit 402 and the prediction unit 404 in the settlement parameter prediction device. In addition, it may also include but is not limited to other module units in the settlement parameter prediction device, which will not be repeated in this example.

[0117] Optionally, the transmission device 506 is used to receive or send data via a network. Specific examples of the network may include wired networks and wireless networks. In one example, the transmission device 506 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers via a network cable so as to communicate with the Internet or a local area network. In one example, the transmission device 506 is a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0118] In addition, the above-mentioned electronic device also includes: a display 508 for displaying the input period attribute data and the output target settlement parameters; and a connection bus 510 for connecting the various module components in the above-mentioned electronic device.

[0119] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0120] According to one aspect of the present application, a computer program product is provided. The computer program product includes a computer program / instruction. The computer program / instruction contains a program code for executing the method shown in the flowchart.

[0121] According to one aspect of the present application, a computer-readable storage medium is provided, and a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the settlement parameter prediction method provided in the above-mentioned various optional implementations.

[0122] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0123] S1, obtaining period attribute data of a target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to a target duration threshold;

[0124] S2, in the settlement parameter prediction network, based on the settlement timing features extracted from the period attribute data, predict the target settlement parameters corresponding to the target settlement period, wherein the settlement parameter prediction network includes a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained multiple times using settlement data samples until the convergence condition is reached, and the convergence condition indicates that the training loss value output by the settlement parameter prediction network in training has reached the set loss condition value.

[0125] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing hardware related to the terminal device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0126] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0127] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0128] In the several embodiments provided in the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0129] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0130] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0131] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for predicting settlement parameters, characterized in that: include: Acquire period attribute data of a target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to a target duration threshold; In the settlement parameter prediction network, based on the settlement timing characteristics extracted from the period attribute data, the target settlement parameters corresponding to the target settlement period are predicted, wherein the settlement parameter prediction network includes a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained multiple times using settlement data samples until a convergence condition is reached, and the convergence condition indicates that the training loss value output by the settlement parameter prediction network during training has reached a set loss condition value.

2. The method according to claim 1, characterized in that The step of obtaining the period attribute data of the target settlement period to be predicted includes: Acquire meteorological data of the target settlement period, and electricity consumption disclosure conditions configured for the electric equipment in the target settlement period, wherein the meteorological data is data collected in the target settlement period to reflect natural weather changes, and the electricity consumption disclosure conditions include: the load of the electric equipment, the capacity information of the electric equipment, and the electricity consumption strategy of the electric equipment; Based on the meteorological data and the electricity consumption disclosure conditions, the period attribute data matching the target settlement period is determined.

3. The method according to claim 1, characterized in that In the settlement parameter prediction network, based on the settlement timing characteristics extracted from the period attribute data, the target settlement parameters corresponding to the target settlement period are predicted, including: Extracting the settlement timing characteristics corresponding to the target settlement period from the period attribute data in the settlement parameter prediction network; In the settlement parameter prediction network, the target settlement parameters of the target settlement period are predicted based on the settlement timing characteristics and the historical settlement timing characteristics corresponding to the historical settlement periods.

4. The method according to claim 1, characterized in that Before obtaining the period attribute data of the target settlement period to be predicted, the method further includes: Inputting the settlement data sample into the initialized settlement parameter prediction network for training to obtain a training result; Input the current label settlement parameter corresponding to the currently input settlement data sample and the currently output training result into the loss function to obtain the current training loss value; When the current training loss value has not yet reached the set loss condition value, adjusting the network parameters in the settlement parameter prediction network in training; When the current training loss value reaches the set loss condition value, it is determined that the settlement parameter prediction network in training has reached the convergence condition.

5. The method according to claim 4, characterized in that Before obtaining the period attribute data of the target settlement period to be predicted, the method further includes: Acquire historical settlement data obtained in a historical settlement period, historical electricity consumption disclosure conditions configured for electric equipment in the historical settlement period, and reference meteorological data of a reference settlement period associated with the historical settlement period, wherein the reference settlement period is a forecast settlement period associated with the corresponding historical settlement period; The historical settlement data, the historical electricity consumption disclosure conditions and the reference meteorological data are normalized to obtain a plurality of input data sets, and the plurality of input data sets are used as the settlement data samples.

6. The method according to claim 4, characterized in that Before obtaining the period attribute data of the target settlement period to be predicted, the method further includes: The data segmentation subnetwork, the double-layer convolution subnetwork and the double-layer feedforward subnetwork are associated and connected to construct the settlement parameter prediction network, wherein the data segmentation subnetwork is used to perform block processing on the input settlement data samples, the double-layer convolution subnetwork includes: a deep convolution subnetwork and a point convolution subnetwork, and the double-layer feedforward subnetwork is used for regression processing.

7. The method according to claim 6, characterized in that The inputting the settlement data sample into the initialized settlement parameter prediction network for training to obtain the training result comprises: In the data segmentation subnetwork, the currently input settlement data sample is processed into blocks to obtain a plurality of sample data blocks that meet the block conditions; Acquire the current sample settlement timing features of the multiple sample data blocks through the double-layer convolution subnetwork, wherein the current sample settlement timing features include: the intra-block timing relationship features determined by the deep convolution subnetwork, and the inter-block timing relationship features determined by the point convolution subnetwork; The current sample settlement time series features are regressed through the double-layer feedforward sub-network to obtain the current output training result.

8. A settlement parameter prediction device, characterized in that: include: A first acquisition unit is used to acquire period attribute data of a target settlement period to be predicted, wherein the period duration of the target settlement period is greater than or equal to a target duration threshold; A prediction unit is used to predict the target settlement parameters corresponding to the target settlement period in the settlement parameter prediction network based on the settlement timing characteristics extracted from the period attribute data, wherein the settlement parameter prediction network includes a two-layer convolutional subnetwork and a two-layer feedforward subnetwork, and the settlement parameter prediction network is a neural network that is iteratively trained multiple times using settlement data samples until a convergence condition is reached, and the convergence condition indicates that the training loss value output by the settlement parameter prediction network in training has reached a set loss condition value.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program is executed by a processor to perform the method described in any one of claims 1 to 7.

10. 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 7 through the computer program.

Images