New energy power price management method and system based on artificial intelligence

Abstract

The application discloses a new energy power price management method and system based on artificial intelligence, and contents include data collection, data expansion, construction of a new energy power price dynamic prediction model and price management. The application relates to the technical field of power price management, and particularly relates to a new energy power price management method and system based on artificial intelligence. The scheme adopts a time sequence convolution module and a deconvolution module, can generate data that is more in line with actual data distribution, combines the authenticity of generated data and the continuity between adjacent time steps to design a loss function, and ensures the quality of generated data. Through design of an output control function and a memory layer structure, nonlinear characteristics of new energy power price data can be more flexibly processed, time sequence dependency and dynamic change characteristics in new energy power price data can be captured, and the prediction accuracy, prediction precision and generalization ability of the model are improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of power price management, in particular to a new energy power price management method and system based on artificial intelligence. BACKGROUND

[0002] The new energy power price management method and system based on artificial intelligence is a system for power price management combined with artificial intelligence technology and data processing technology, and can monitor and analyze the situation of the power market in real time. Through learning and analyzing a large amount of historical data, the power price trend can be accurately predicted.

[0003] The traditional new energy power price management data expansion method relies on simple data replication and interpolation, and these methods have the problems of insufficient data diversity, inability to effectively capture the complex correlation and dynamic changes between data. When constructing a new energy power price dynamic prediction model, the accuracy and reliability of the prediction results are affected by the low data quality; the traditional new energy power price dynamic prediction model relies on simple linear regression, time series analysis and shallow neural network methods, and has the problem of insufficient prediction accuracy when processing complex and nonlinear new energy power price data. SUMMARY

[0004] In view of the above problems, in order to overcome the defects of the prior art, the application provides a new energy power price management method and system based on artificial intelligence. The traditional new energy power price management data expansion method relies on simple data replication and interpolation, and these methods have the problems of insufficient data diversity, inability to effectively capture the complex correlation and dynamic changes between data. When constructing a new energy power price dynamic prediction model, the accuracy and reliability of the prediction results are affected by the low data quality, the time sequence convolution module and the deconvolution module are adopted in the scheme, which can capture the time sequence dependence and spatial features between data, generate new samples more consistent with the actual data distribution, assign attention weights to time steps by designing an attention layer, so that the generated data is more coherent and consistent in time sequence, and the loss function is designed by combining the authenticity of the generated data and the coherence between adjacent time steps, so as to ensure the quality of the generated data; the traditional new energy power price dynamic prediction model relies on simple linear regression, time series analysis and shallow neural network methods, and has the problem of insufficient prediction accuracy when processing complex and nonlinear new energy power price data. The output control function is designed in the scheme, which can more flexibly process the nonlinear characteristics of the new energy power price data, the memory layer structure is designed, which can capture the time sequence dependence and dynamic change characteristics of the new energy power price data, effectively solve the problems of the traditional model when processing time series data, and improve the prediction accuracy and generalization ability of the model.

[0005] The technical scheme adopted by the present application is as follows: a new energy power price management method based on artificial intelligence, which comprises the following steps:

[0006] Step S1: data acquisition;

[0007] Step S2: data expansion;

[0008] Step S3: constructing a new energy power price dynamic prediction model;

[0009] Step S4: price management.

[0010] Further, in step S1, the data acquisition is to collect historical new energy power generation data, meteorological data, power grid state data, cost data and power price; the new energy power generation data includes power generation and power generation power; the power grid state data includes normal state and abnormal state; the cost data includes construction cost and maintenance cost.

[0011] Further, in step S2, the data expansion specifically comprises the following steps:

[0012] Step S21: constructing a generator, specifically comprising the following steps:

[0013] Step S211: generating an embedding vector, represented as follows:

[0014] ;

[0015] Wherein, represents the embedding vector, represents a linear rectification function, and represent noise weight matrix and time weight matrix respectively, represents random noise, v represents time step, represents embedding bias;

[0016] Step S212: constructing a time series convolution module, represented as follows:

[0017] ;

[0018] Wherein, represents the output of the time series convolution module, j represents the index of the convolution kernel element, K represents the convolution kernel size, represents one-dimensional convolution step, represents convolution padding, represents the element of the vector after adjustment by step and padding pad, represents the jth element of the convolution kernel;

[0019] Step S213: constructing a deconvolution module, denoted as follows:

[0020] ;

[0021] wherein, denotes the output of the deconvolution module, j denotes the index of the element of the convolution kernel, denotes the deconvolution kernel size, denotes the one-dimensional deconvolution step size, denotes the deconvolution padding item, denotes the vector is adjusted by the deconvolution step size and the deconvolution padding item ; denotes the jth element of the deconvolution kernel;

[0022] Step S22: constructing a discriminator, specifically comprising the following steps:

[0023] Step S221: constructing a feature learning module, denoted as follows:

[0024] ;

[0025] wherein, denotes the output of the feature learning module, denotes a modified linear function with leakage, u denotes the layer index of the feature learning module, and U denotes the maximum number of layers of the feature learning module, and denote the weight and bias of the u-th layer of the feature learning module, respectively, denotes the input of the u-th layer of the feature learning module;

[0026] Step S222: designing an attention layer, denoted as follows:

[0027] ;

[0028] wherein, denotes the attention weight of the v-th time step, denotes a normalized exponential function, denotes a hyperbolic tangent function, denotes the input of the attention layer, denotes the input weight of the attention layer, denotes the attention weight of the v-1th time step, denotes the adjustment weight of the v-1th time step, denotes the bias of the attention layer;

[0029] Step S23: constructing a loss function, denoted as follows: ​

[0030] ;

[0031] wherein, denotes the loss of the generator, y denotes the input of the generator, denotes the probability distribution of the generator input, denotes the generator, denotes the discriminator, denotes taking the logarithm, denotes taking the expectation value in the parentheses when y is randomly drawn from the distribution ; denotes the loss adjustment coefficient, denotes the square of the L2 norm between the data generated by adjacent time steps v and v-1, denotes the divergence loss adjustment coefficient, denotes the distribution of the data generated by the generator, denotes the KL divergence between the distribution of the data generated by the generator and the real data distribution, denotes the loss of the discriminator, denotes the real data, denotes the probability distribution of the real data, denotes taking the expectation value in the parentheses when is randomly drawn from the distribution ;

[0032] Step S24: Data expansion, expand the original data by generating new energy power generation data, meteorological data, power grid state data, cost data and power price data through the generator.

[0033] Further, in step S3, the new energy power price dynamic prediction model is constructed, specifically including the following steps:

[0034] Step S31: Design the output control function, denoted as follows:

[0035] ;

[0036] wherein, denotes the input value of the output control function, denotes the output value of the output control function, denotes the sigmoid function, denotes the natural constant based logarithmic function, denotes the inverse tangent function, denotes square root;

[0037] Step S32: Design the memory layer, specifically including the following steps:

[0038] Step S321: design an input unit, denoted as follows:

[0039] ;

[0040] wherein b denotes the index of the time point of the memory layer, denotes the output value of the input unit at the b time point, and denote the weight and bias of the input unit, respectively, denotes the state factor of the memory layer at the b-1 time point, denotes the input value of the input unit at the b time point;

[0041] Step S322: design a forgetting unit, denoted as follows:

[0042] ;

[0043] wherein denotes the output value of the forgetting unit at the b time point, and denote the weight and bias of the forgetting unit, respectively, denotes the input value of the forgetting unit at the b time point;

[0044] Step S323: construct an initial state unit, denoted as follows:

[0045] ;

[0046] wherein denotes the initial state factor at the b time point, and denote the initial state weight and initial state bias, respectively, denotes the input value of the initial state unit at the b time point;

[0047] Step S324: design an update unit, denoted as follows:

[0048] ;

[0049] wherein denotes the output value of the update unit at the b time point, and denote the weight and bias of the update unit, respectively, denotes the input value of the update unit at the b time point;

[0050] Step S325: design an output unit, denoted as follows:

[0051] ;

[0052] wherein represents the output value of the output unit at the b time point, respectively represent the weight and bias of the output unit, represents the input value of the output unit at the b time point;

[0053] Step S326: output the state factor, which is represented as follows:

[0054]

[0055] wherein, represents the state factor of the memory layer at the b time point;

[0056] Step S33: design the loss function, which is represented as follows:

[0057]

[0058] wherein, represents the loss value of the model, n represents the index of the data used for model training, and N represents the total number of data used for model training, represents the true value of the nth data, represents the model prediction value of the nth data, represents taking the absolute value;

[0059] Step S34: set the model label, and set the power price as the label data of the model.

[0060] Further, in step S4, the price management is performed by collecting new energy power generation data, meteorological data, power grid state data and cost data, inputting the data into a new energy power price dynamic prediction model, and the model predicts the power price. The power price predicted by the model is adjusted to the real-time power price.

[0061] The present application provides a new energy power price management method and system based on artificial intelligence, which comprises a data acquisition module, a data expansion module, a new energy power price dynamic prediction model construction module and a price management module.

[0062] The data acquisition module acquires historical new energy power generation data, meteorological data, power grid state data, cost data and power price, and sends the data to the data expansion module.

[0063] The data expansion module receives the data sent by the data acquisition module, expands the original data by constructing a generator, and sends the data to the new energy power price dynamic prediction model construction module.

[0064] ​​​The new energy power price dynamic prediction model construction module receives the data sent by the data expansion module, constructs a new energy power price dynamic prediction model, and sends the data to the price management module;

[0065] The price management module receives the data sent by the new energy power price dynamic prediction model construction module, predicts the power price by using the new energy power price dynamic prediction model, and adjusts the power price predicted by the model to the real-time power price.

[0066] The beneficial effects achieved by the above-mentioned scheme are as follows:

[0067] (1) The traditional new energy power price management data expansion method relies on simple data replication and interpolation, which has the problems of insufficient data diversity, inability to effectively capture the complex correlation and dynamic changes between data, and the like. When constructing a new energy power price dynamic prediction model, the data quality is not high, which affects the accuracy and reliability of the prediction results. The time series convolution module and the deconvolution module are used in the present scheme, which can capture the time series dependence and spatial features between data, generate new samples that are more consistent with the actual data distribution, and allocate attention weights to time steps by designing an attention layer, so that the generated data is more coherent and consistent in time series. The loss function is designed by combining the authenticity of the generated data and the coherence between adjacent time steps, which ensures the quality of the generated data.

[0068] (2) The traditional new energy power price dynamic prediction model relies on simple linear regression, time series analysis and shallow neural network methods, which have the problem of insufficient prediction accuracy when processing complex and nonlinear new energy power price data. The output control function is designed in the present scheme, which can more flexibly process the nonlinear features of new energy power price data. The memory layer structure is designed, which can capture the time series dependence and dynamic change characteristics of new energy power price data, effectively solving the problems of traditional models when processing time series data, and improving the prediction accuracy and generalization ability of the model. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1 A schematic diagram of the new energy power price management method based on artificial intelligence provided by the present application;

[0070] Figure 2 A schematic diagram of the new energy power price management system based on artificial intelligence provided by the present application;

[0071] Figure 3 A schematic diagram of data expansion;

[0072] Figure 4 A schematic diagram of constructing a new energy power price dynamic prediction model.

[0073] The accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of the specification, illustrate embodiments of the application, and are used to explain the application, but are not intended to limit the application. DETAILED DESCRIPTION

[0074] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0075] In the description of the application, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the application.

[0076] Embodiment one, refer to Figure 1 The application provides a new energy power price management method based on artificial intelligence, which comprises the following steps:

[0077] Step S1: data acquisition, collecting historical new energy power generation data, meteorological data, power grid state data, cost data and power price; the new energy power generation data includes power generation and power generation power; the power grid state data includes normal state and abnormal state; the cost data includes construction cost and maintenance cost;

[0078] Step S2: data expansion, data expansion is performed by constructing a generator, a discriminator and a loss function;

[0079] Step S3: constructing a new energy power price dynamic prediction model, the new energy power price dynamic prediction model is constructed by designing an output control function, a memory layer and a loss function;

[0080] Step S4: price management, the power price predicted by the new energy power price dynamic prediction model is adjusted to the real-time power price.

[0081] Embodiment two, refer to Figure 1 and Figure 3 This embodiment is based on the above-mentioned embodiment, and the data expansion specifically comprises the following steps:

[0082] Step S21: constructing a generator, specifically comprising the following steps:

[0083] Step S211: Generate an embedding vector, embedding the random noise and time step into a high-dimensional space, as shown below:

[0084] ;

[0085] in, Represents an embedding vector. Represents the linear rectified function. and These represent the noise weight matrix and the time weight matrix, respectively. This represents random noise, and v represents the time step. Indicates embedding bias;

[0086] Step S212: Construct the temporal convolution module, as shown below:

[0087] ;

[0088] in, This represents the output of the temporal convolution module, where j represents the index of the convolution kernel element, and K represents the kernel size. This represents the stride of a one-dimensional convolution. Indicates convolution padding terms. Representing vectors After step length And the element after the padding is adjusted. This represents the j-th element of the convolution kernel;

[0089] Step S213: Construct the deconvolution module, as shown below:

[0090] ;

[0091] in, This represents the output of the deconvolution module, where j represents the index of the convolution kernel element. Indicates the size of the deconvolution kernel. This represents the stride of a one-dimensional deconvolution. Indicates the deconvolution padding term. Representing vectors After deconvolution stride and deconvolution padding Adjusted elements, The deconvolution kernel is represented by the first... One element;

[0092] Step S22: Construct the discriminator, which specifically includes the following steps:

[0093] Step S221: constructing a feature learning module, adopting a multi-layer perceptron structure to design the feature learning module of the discriminator, represented as follows:

[0094] ;

[0095] wherein, represents the output of the feature learning module, represents a modified linear function with leakage, u represents the layer index of the feature learning module, and U represents the maximum number of layers of the feature learning module, and respectively represent the weight and bias of the u-th layer of the feature learning module, represents the input of the u-th layer of the feature learning module.

[0096] Step S222: designing an attention layer, assigning attention weights to time steps, represented as follows:

[0097] ;

[0098] wherein, represents the attention weight of the v-th time step, represents a normalized exponential function, represents a hyperbolic tangent function, represents the input of the attention layer, represents the input weight of the attention layer, represents the attention weight of the v-1-th time step, represents the adjustment weight of the v-1-th time step, represents the bias of the attention layer.

[0099] Step S23: constructing a loss function, represented as follows:

[0100] ;

[0101] wherein, represents the loss of the generator, y represents the input of the generator, represents the probability distribution of the generator input, represents the generator, represents the discriminator, represents taking the logarithm, represents taking the expectation value inside the parentheses when y is randomly drawn from the distribution , represents a loss adjustment coefficient, represents the square of the L2 norm between the data generated by adjacent time steps v and v-1, represents a divergence loss adjustment coefficient, represents the distribution of the data generated by the generator, represents the KL divergence between the distribution of the data generated by the generator and the real data distribution, represents the loss of the discriminator, represents the real data, represents the probability distribution of the real data, represents when the expectation value in the parentheses is taken;

[0102] Step S24: Data expansion, new energy power generation data, meteorological data, grid state data, cost data and power price data are generated by the generator to expand the original data.

[0103] By performing the above operation, the traditional new energy power price management data expansion method relies on simple data replication and interpolation, which has the problems of insufficient data diversity, inability to effectively capture the complex correlation and dynamic changes between data. Lead to the construction of new energy power price dynamic prediction model, because of the low data quality, affect the accuracy and reliability of the prediction result, this scheme adopts time convolution module and deconvolution module, which can capture the time sequence dependence and spatial features between data, generate new samples more consistent with the actual data distribution, assign attention weights to time steps by designing attention layers, make the generated data more coherent and consistent in time sequence, combine the authenticity of generated data and the coherence between adjacent time steps to design the loss function, ensure the quality of generated data.

[0104] Embodiment Three, refer to Figure 1 and Figure 4 , this embodiment is based on the above embodiment, the construction of new energy power price dynamic prediction model, specifically includes the following steps:

[0105] Step S31: Design the output control function, as follows:

[0106] ;

[0107] wherein, represents the input value of the output control function, represents the output value of the output control function, represents the sigmoid function, represents the natural constant based logarithmic function, represents the inverse tangent function, represents the square root;

[0108] Step S32: Design the memory layer, specifically including the following steps:

[0109] Step S321: Design the input unit, as follows:​

[0110] ;

[0111] wherein b denotes an index of a time point of a memory layer, denotes an output value of an input unit at the b time point, and denote a weight and a bias of the input unit, respectively, denotes a state factor of the memory layer at the b-1 time point, denotes an input value of the input unit at the b time point;

[0112] Step S322: designing a forget unit, denoted as follows:

[0113] ;

[0114] wherein denotes an output value of the forget unit at the b time point, and denote a weight and a bias of the forget unit, respectively, denotes an input value of the forget unit at the b time point;

[0115] Step S323: constructing an initial state unit, denoted as follows:

[0116] ;

[0117] wherein denotes an initial state factor at the b time point, and denote an initial state weight and an initial state bias, respectively, denotes an input value of the initial state unit at the b time point;

[0118] Step S324: designing an update unit, denoted as follows:

[0119] ;

[0120] wherein denotes an output value of the update unit at the b time point, and denote a weight and a bias of the update unit, respectively, denotes an input value of the update unit at the b time point;

[0121] Step S325: designing an output unit, denoted as follows:

[0122] ;

[0123] wherein denotes the output value of the output unit at the b-th time point, and denote the weight and bias of the output unit, respectively, denotes the input value of the output unit at the b-th time point;

[0124] Step S326: output the state factor, denoted as follows:

[0125] ;

[0126] wherein, denotes the state factor of the memory layer at the b-th time point;

[0127] Step S33: design the loss function, denoted as follows:

[0128] ;

[0129] wherein, denotes the loss value of the model, n denotes the index of the data used for model training, and N denotes the total number of data used for model training, denotes the true value of the n-th data, denotes the model prediction value of the n-th data, denotes taking the absolute value;

[0130] Step S34: set the model label, and set the electricity price as the label data of the model.

[0131] By performing the above operation, for the traditional new energy power price dynamic prediction model, the method relies on simple linear regression, time series analysis and shallow neural network, and there is a problem of insufficient prediction accuracy when processing complex and nonlinear new energy power price data. The present scheme can more flexibly process the nonlinear characteristics of new energy power price data by designing an output control function, and can capture the time sequence dependence and dynamic change characteristics in new energy power price data by designing a memory layer structure, effectively solving the problems of traditional models when processing time series data, and improving the prediction accuracy and generalization ability of the model.

[0132] Embodiment four, refer to Figure 1 This embodiment is based on the above-mentioned embodiment, and the price management is performed by collecting new energy power generation data, meteorological data, power grid state data and cost data, inputting the data into a new energy power price dynamic prediction model, and adjusting the model predicted power price to the real-time power price.

[0133] Embodiment five, refer to Figure 1 and Figure 2The embodiment is based on the above embodiment, and the new energy power price management method and system based on artificial intelligence provided by the application comprises a data acquisition module, a data expansion module, a new energy power price dynamic prediction model construction module and a price management module;

[0134] The data acquisition module acquires historical new energy power generation data, meteorological data, power grid state data, cost data and power prices, and sends the data to the data expansion module;

[0135] The data expansion module receives the data sent by the data acquisition module, expands the original data by constructing a generator, and sends the data to the new energy power price dynamic prediction model construction module;

[0136] The new energy power price dynamic prediction model construction module receives the data sent by the data expansion module, constructs a new energy power price dynamic prediction model, and sends the data to the price management module;

[0137] The price management module receives the data sent by the new energy power price dynamic prediction model construction module, predicts the power price by using the new energy power price dynamic prediction model, and adjusts the power price predicted by the model to the real-time power price.

[0138] It should be noted that in this paper, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0139] Although the embodiments of the application have been shown and described, it can be understood by those of ordinary skill in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the application.

[0140] The above describes the application and its embodiments, which are not limited, and the drawings only show one of the embodiments of the application, and the actual structure is not limited thereto. In general, if those of ordinary skill in the art are inspired, without departing from the principles and spirits of the application, without creative design, similar structure and embodiments of the technical solutions can be designed, which should belong to the protection scope of the application.

Claims

1. An artificial intelligence-based new energy power price management method, characterized in that: The method comprises the following steps: Step S1: data collection, collecting historical new energy power generation data, meteorological data, power grid state data, cost data and electricity price; the new energy power generation data includes power generation and power generation power; the power grid state data includes normal state and abnormal state; the cost data includes construction cost and maintenance cost; Step S2: data expansion, data expansion is performed by constructing a generator, a discriminator and a loss function; Wherein, the generator is constructed, specifically including the following steps: Step S211: generate an embedding vector, represented as follows: ； wherein, denotes an embedding vector, denotes a linear rectification function, and denote a noise weight matrix and a time weight matrix, respectively, denotes a random noise, v denotes a time step, denotes an embedding bias; Step S212: construct a time series convolution module, represented as follows: ； wherein, represents the output of the temporal convolution module, j represents the index of the convolution kernel element, K represents the convolution kernel size, represents the one-dimensional convolution step size, represents the convolution padding item, represents the vector the element adjusted by the step size and the padding item pad, represents the jth element of the convolution kernel; Step S213: construct a deconvolution module, represented as follows: ； in, This represents the output of the deconvolution module, where j represents the index of the convolution kernel element. Indicates the size of the deconvolution kernel. This represents the stride of a one-dimensional deconvolution. Indicates the deconvolution padding term. Representing vectors After deconvolution stride and deconvolution padding Adjusted elements, The deconvolution kernel is represented by the first... One element; Step S3: construct a new energy power price dynamic prediction model, design an output control function, a memory layer and a loss function to construct a new energy power price dynamic prediction model; Step S31: design an output control function, represented as follows: ； wherein, represents an input value of an output control function, represents an output value of an output control function, represents a sigmoid function, represents a natural constant-based logarithmic function, represents an inverse tangent function, represents a square root; Step S32: design a memory layer, specifically including the following steps: Step S321: design an input unit, represented as follows: ； wherein b denotes an index of a time point of the memory layer, denotes an output value of the input unit at the b time point, and denote a weight and a bias of the input unit, respectively, denotes a state factor of the memory layer at the b−1 time point, denotes an input value of the input unit at the b time point; Step S322: design a forgetting unit, represented as follows: ； wherein, represents an output value of the forget gate at the b-th time point, and respectively represent a weight and a bias of the forget gate, represents an input value of the forget gate at the b-th time point; Step S323: construct an initial state unit, represented as follows: ； wherein, represents an initial state factor at a time point b, and respectively represent an initial state weight and an initial state bias, represents an input value of the initial state unit at a time point b; Step S324: design an update unit, represented as follows: ； wherein, represents the output value of the update unit at the b-th time point, and respectively represent the weight and the bias of the update unit, represents the input value of the update unit at the b-th time point; Step S325: design an output unit, represented as follows: ； wherein, represents an output value of the output unit at a time point b, and respectively represent a weight and a bias of the output unit, represents an input value of the output unit at a time point b; Step S326: output state factors, represented as follows: ； wherein denotes the state factor of the memory layer at time b; Step S33: design a loss function, represented as follows: ； wherein, represents a loss value of the model, n represents an index of data used for model training, N represents a total number of data used for model training, represents a true value of the nth data, represents a model predicted value of the nth data, represents taking an absolute value; Step S34: set the model label, set the electricity price as the label data of the model; Step S4: price management, predict the electricity price by using the new energy power price dynamic prediction model, and adjust the electricity price predicted by the model to the real-time electricity price. 2.The new energy power price management method based on artificial intelligence according to claim 1, characterized in that: In step S2, the data expansion specifically includes the following steps: Step S21: construct a generator; Step S22: construct a discriminator, specifically including the following steps: Step S221: construct a feature learning module, represented as follows: ； wherein, represents an output of the feature learning module, represents a modified linear function with leakage, u represents a layer index of the feature learning module, and U represents a maximum number of layers of the feature learning module, and respectively represent a weight and a bias of the u-th layer of the feature learning module, represents an input of the u-th layer of the feature learning module; Step S222: design an attention layer, represented as follows: ； wherein, denotes the attention weight for the v-th time step, denotes the normalized exponential function, denotes the hyperbolic tangent function, denotes the input of the attention layer, denotes the input weight of the attention layer, denotes the attention weight for the v-1 -th time step, denotes the adjustment weight for the v-1 -th time step, denotes the bias of the attention layer; Step S23: construct a loss function, represented as follows: ； wherein, represents the loss of the generator, y represents the input of the generator, represents the probability distribution of the generator input, represents the generator, represents the discriminator, represents taking the logarithm, represents taking the expectation value in the brackets when y is randomly drawn from the distribution , represents the loss adjustment coefficient, represents the square of the L2 norm between the data generated by adjacent time steps v and v-1, represents the divergence loss adjustment coefficient, represents the distribution of the data generated by the generator, represents the KL divergence between the distribution of the data generated by the generator and the real data distribution, represents the loss of the discriminator, represents the real data, represents the probability distribution of the real data, represents taking the expectation value in the brackets when is randomly drawn from the distribution ; Step S24: data expansion, expand the original data by generating new energy power generation data, meteorological data, power grid state data, cost data and electricity price data through the generator. 3.The new energy power price management method based on artificial intelligence according to claim 1, characterized in that: In step S1, the data collection is to collect historical new energy power generation data, meteorological data, power grid state data, cost data and electricity price; the new energy power generation data includes power generation and power generation power; the power grid state data includes normal state and abnormal state; the cost data includes construction cost and maintenance cost; In step S4, the price management is to collect new energy power generation data, meteorological data, power grid state data and cost data, input the data into the new energy power price dynamic prediction model, and predict the electricity price by the model, and adjust the electricity price predicted by the model to the real-time electricity price.

4. The new energy power price management system based on artificial intelligence, used to realize the new energy power price management method based on artificial intelligence as claimed in any one of claims 1-3, characterized in that: It comprises a data collection module, a data expansion module, a new energy power price dynamic prediction model construction module and a price management module; The data collection module collects historical new energy power generation data, meteorological data, power grid state data, cost data and power price, and sends the data to a data expansion module; The data expansion module receives the data sent by the data collection module, expands the original data by constructing a generator, and sends the data to a new energy power price dynamic prediction model construction module; The new energy power price dynamic prediction model construction module receives the data sent by the data expansion module, constructs a new energy power price dynamic prediction model, and sends the data to a price management module; The price management module receives the data sent by the new energy power price dynamic prediction model construction module, predicts the power price by using the new energy power price dynamic prediction model, and adjusts the power price predicted by the model to a real-time power price.

Images