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

Abstract

The invention discloses a new energy electric power price management method and system based on artificial intelligence. The method comprises the steps of data acquisition, data expansion, construction of a new energy electric power price dynamic prediction model and price management. The invention 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, and the method employs a time sequence convolution module and a deconvolution module, can generate data which better accords with the actual data distribution, and improves the efficiency of power price management. A loss function is designed by combining the authenticity of the generated data and the coherence between the generated data of adjacent time steps, so that the quality of the generated data is ensured; by designing an output control function and a memory layer structure, nonlinear features of the new energy power price data can be processed more flexibly, time sequence dependence and dynamic change features in the new energy power price data are captured, and prediction accuracy, prediction precision and generalization ability of the model are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electricity price management, and specifically to a new energy electricity price management method and system based on artificial intelligence. Background Art

[0002] The new energy power price management method and system based on artificial intelligence is a system that combines artificial intelligence technology and data processing technology to manage power prices. It can monitor and analyze the power market in real time. By learning and analyzing a large amount of historical data, the trend of power prices can be accurately predicted.

[0003] Traditional data expansion methods for new energy power price management rely on simple data replication and interpolation. These methods have the problem of insufficient data diversity and inability to effectively capture the complex correlations and dynamic changes between data. As a result, when constructing a dynamic prediction model for new energy power prices, the accuracy and reliability of the prediction results are affected due to low data quality; traditional dynamic prediction models for new energy power prices rely on simple linear regression, time series analysis, and shallow neural network methods, which have the problem of insufficient prediction accuracy when dealing with complex and nonlinear new energy power price data. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a new energy electricity price management method and system based on artificial intelligence. The traditional new energy electricity price management data expansion method relies on simple data replication and interpolation. These methods have the problems of insufficient data diversity and inability to effectively capture the complex correlations and dynamic changes between data. As a result, when constructing a dynamic prediction model for renewable energy power prices, the accuracy and reliability of the prediction results are affected by the low data quality. This solution adopts a time series convolution module and a deconvolution module, which can capture the time series dependency and spatial characteristics between data, generate new samples that are more in line with the actual data distribution, and design an attention layer to assign attention weights to time steps, 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 the generated data of adjacent time steps to ensure the quality of the generated data. The traditional dynamic prediction model for renewable energy power prices 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 renewable energy power price data. This solution can more flexibly process the nonlinear characteristics of renewable energy power price data by designing an output control function, and can capture the time series dependency and dynamic change characteristics in renewable energy power price data by designing a memory layer structure, effectively solving the problems existing in traditional models when processing time series data, and improving the prediction accuracy and generalization ability of the model.

[0005] The technical solution adopted by the present invention is as follows: a new energy power price management method based on artificial intelligence, the method comprising the following steps:

[0006] Step S1: data collection;

[0007] Step S2: data expansion;

[0008] Step S3: construct a dynamic prediction model for new energy power prices;

[0009] Step S4: Price management.

[0010] Furthermore, in step S1, the data collection is to collect historical new energy power generation data, meteorological data, grid status data, cost data and electricity prices; the new energy power generation data includes power generation and power generation power; the grid status data includes normal state and abnormal state; the cost data includes construction cost and maintenance cost.

[0011] Furthermore, in step S2, the data expansion specifically includes the following steps:

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

[0013] Step S211: Generate an embedding vector, which is expressed as follows:

[0014] ;

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

[0016] Step S212: construct a temporal convolution module, which is expressed as follows:

[0017] ;

[0018] in, 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, represents the convolution padding term, Representation vector Passing step length and the padding item after the element is adjusted. Represents the jth element of the convolution kernel;

[0019] Step S213: construct a deconvolution module, which is expressed as follows:

[0020] ;

[0021] in, represents the output of the deconvolution module, j represents the index of the convolution kernel element, represents the deconvolution kernel size, represents the one-dimensional deconvolution step size, represents the deconvolution padding term, Representation vector After deconvolution step and deconvolution padding The adjusted elements, represents the deconvolution kernel elements;

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

[0023] Step S221: construct a feature learning module, which is expressed as follows:

[0024] ;

[0025] in, represents the output of the feature learning module, represents the rectified linear function with leakage, u represents the layer index of the feature learning module, 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;

[0026] Step S222: Design the attention layer, which is expressed as follows:

[0027] ;

[0028] in, represents the attention weight of the vth time step, represents the normalized exponential function, represents the 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-1th time step, represents the adjustment weight of the v-1th time step, represents the bias of the attention layer;

[0029] Step S23: construct a loss function, which is expressed as follows:

[0030] ;

[0031] in, represents the loss of the generator, y represents the input of the generator, represents the probability distribution of the generator input, represents a generator, represents the discriminator, represents taking the logarithm, It means that when y is distributed from When randomly selecting from , take the expected value in brackets. 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 distribution of the real data, represents the loss of the discriminator, Represents real data, represents the probability distribution of real data, Indicates when From the distribution When randomly selecting from , take the expected value in brackets;

[0032] Step S24: Data expansion, by using a generator to generate new energy power generation data, meteorological data, grid status data, cost data and electricity price data to expand the original data.

[0033] Furthermore, in step S3, the construction of a new energy power price dynamic prediction model specifically includes the following steps:

[0034] Step S31: Design an output control function, which is expressed as follows:

[0035] ;

[0036] in, represents the input value of the output control function, Represents the output value of the output control function, represents the sigmoid function, represents the logarithmic function with a natural constant as base, represents the inverse tangent function, Indicates square root;

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

[0038] Step S321: Design an input unit, as shown below:

[0039] ;

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

[0041] Step S322: Design a forgetting unit, which is expressed as follows:

[0042] ;

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

[0044] Step S323: construct an initial state unit, which is expressed as follows:

[0045] ;

[0046] in, represents the initial state factor at time point b, and They represent the initial state weight and initial state bias respectively, Represents the input value of the initial state unit at time point b;

[0047] Step S324: Design update unit, which is expressed as follows:

[0048] ;

[0049] in, represents the output value of the update unit at time point b, and Represent the weight and bias of the update unit respectively, represents the input value of the update unit at time point b;

[0050] Step S325: Design the output unit, as shown below:

[0051] ;

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

[0053] Step S326: Output the state factor, which is expressed as follows:

[0054] ;

[0055] in, Represents the state factor of the memory layer at time b;

[0056] Step S33: Design a loss function, which is expressed as follows:

[0057] ;

[0058] in, 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, Indicates taking the absolute value;

[0059] Step S34: Setting the model label, and setting the electricity price as the label data of the model.

[0060] Furthermore, in step S4, the price management is carried out by collecting new energy power generation data, meteorological data, grid status data and cost data, inputting the data into a dynamic prediction model for new energy electricity prices, the model predicts the electricity price, and adjusting the electricity price predicted by the model to the real-time electricity price.

[0061] The artificial intelligence-based new energy power price management method and system provided by the present invention include a data acquisition module, a data expansion module, a module for building a dynamic prediction model for new energy power prices, and a price management module;

[0062] The data acquisition module collects historical renewable energy power generation data, meteorological data, grid status data, cost data and electricity prices, 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 module for constructing a dynamic prediction model for new energy power prices;

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

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

[0066] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0067] (1) Traditional methods for expanding data for new energy power price management rely on simple data replication and interpolation. These methods have the problem of insufficient data diversity and inability to effectively capture the complex correlations and dynamic changes between data. As a result, when constructing a dynamic prediction model for new energy power prices, the accuracy and reliability of the prediction results are affected by the low data quality. This scheme uses a temporal convolution module and a deconvolution module to capture the temporal dependencies and spatial features between data and generate new samples that are more in line with the actual data distribution. By designing an attention layer to assign attention weights to time steps, the generated data is more coherent and consistent in time. The loss function is designed by combining the authenticity of the generated data and the coherence between the generated data of adjacent time steps to ensure the quality of the generated data.

[0068] (2) The traditional dynamic prediction model of renewable energy power price relies on simple linear regression, time series analysis and shallow neural network methods. When dealing with complex and nonlinear renewable energy power price data, the prediction accuracy is insufficient. This scheme can handle the nonlinear characteristics of renewable energy power price data more flexibly by designing the output control function. By designing the memory layer structure, it can capture the time series dependence and dynamic change characteristics in renewable energy power price data, effectively solving the problems existing in traditional models when dealing with time series data and improving the prediction accuracy and generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

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

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

[0071] Figure 3 Schematic diagram of data expansion;

[0072] Figure 4 Schematic diagram of constructing a dynamic prediction model for new energy electricity prices.

[0073] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0074] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0075] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0076] Example 1, see Figure 1 The present invention provides a new energy power price management method based on artificial intelligence, which comprises the following steps:

[0077] Step S1: Data collection, collecting historical new energy power generation data, meteorological data, grid status data, cost data and electricity prices; the new energy power generation data includes power generation and power generation; the grid status 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: construct a new energy power price dynamic prediction model by designing an output control function, a memory layer and a loss function;

[0080] Step S4: Price management: Use the new energy power price dynamic prediction model to predict the power price, and adjust the power price predicted by the model to the real-time power price.

[0081] Example 2, see Figure 1 and Figure 3 This embodiment is based on the above embodiment, and the data expansion specifically includes the following steps:

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

[0083] Step S211: Generate an embedding vector to embed random noise and time steps into a high-dimensional space, expressed as follows:

[0084] ;

[0085] in, represents the embedding vector, represents the linear rectification function, and denote the noise weight matrix and the time weight matrix respectively, represents random noise, v represents the time step, represents the embedding bias;

[0086] Step S212: construct a temporal convolution module, which is expressed as follows:

[0087] ;

[0088] in, 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, represents the convolution padding term, Representation vector Passing step length and the padding item after the element is adjusted. Represents the jth element of the convolution kernel;

[0089] Step S213: construct a deconvolution module, which is expressed as follows:

[0090] ;

[0091] in, represents the output of the deconvolution module, j represents the index of the convolution kernel element, represents the deconvolution kernel size, represents the one-dimensional deconvolution step size, represents the deconvolution padding term, Representation vector After deconvolution step and deconvolution padding The adjusted elements, represents the deconvolution kernel elements;

[0092] Step S22: constructing a discriminator, specifically including the following steps:

[0093] Step S221: construct a feature learning module, and use the structure of a multi-layer perceptron to design the feature learning module of the discriminator, which is expressed as follows:

[0094] ;

[0095] in, represents the output of the feature learning module, represents the rectified linear function with leakage, u represents the layer index of the feature learning module, 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: Design an attention layer to assign attention weights to time steps, expressed as follows:

[0097] ;

[0098] in, represents the attention weight of the vth time step, represents the normalized exponential function, represents the 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-1th time step, represents the adjustment weight of the v-1th time step, represents the bias of the attention layer;

[0099] Step S23: construct a loss function, which is expressed as follows:

[0100] ;

[0101] in, represents the loss of the generator, y represents the input of the generator, represents the probability distribution of the generator input, represents a generator, represents the discriminator, represents taking the logarithm, It means that when y is distributed from When randomly selecting from , take the expected value in brackets. 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 distribution of the real data, represents the loss of the discriminator, Represents real data, represents the probability distribution of real data, Indicates when From the distribution When randomly selecting from , take the expected value in brackets;

[0102] Step S24: Data expansion, by using a generator to generate new energy power generation data, meteorological data, grid status data, cost data and electricity price data to expand the original data.

[0103] By performing the above operations, the traditional data expansion methods for new energy power price management rely on simple data duplication and interpolation. These methods have the problem of insufficient data diversity and inability to effectively capture the complex correlation and dynamic changes between data. As a result, when constructing a dynamic prediction model for new energy power prices, the accuracy and reliability of the prediction results are affected by the low data quality. This solution uses a temporal convolution module and a deconvolution module, which can capture the temporal dependencies and spatial characteristics between data, and generate new samples that are more in line with the actual data distribution. By designing an attention layer to assign attention weights to time steps, the generated data is more coherent and consistent in time. The loss function is designed by combining the authenticity of the generated data and the coherence between the generated data of adjacent time steps to ensure the quality of the generated data.

[0104] Example 3, see Figure 1 and Figure 4 This embodiment is based on the above embodiment, and the construction of a new energy power price dynamic prediction model specifically includes the following steps:

[0105] Step S31: Design an output control function, which is expressed as follows:

[0106] ;

[0107] in, represents the input value of the output control function, Represents the output value of the output control function, represents the sigmoid function, represents the logarithmic function with a natural constant as base, represents the inverse tangent function, Indicates square root;

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

[0109] Step S321: Design an input unit, as shown below:

[0110] ;

[0111] Among them, b represents the index of the time point of the memory layer, represents the output value of the input unit at time point b, and denote the weight and bias of the input unit respectively, represents the state factor of the memory layer at time b-1, represents the input value of the input unit at time point b;

[0112] Step S322: Design a forgetting unit, which is expressed as follows:

[0113] ;

[0114] in, represents the output value of the forgetting unit at time point b, and denote the weight and bias of the forgetting unit respectively, represents the input value of the forgetting unit at time point b;

[0115] Step S323: construct an initial state unit, which is expressed as follows:

[0116] ;

[0117] in, represents the initial state factor at time point b, and They represent the initial state weight and initial state bias respectively, Represents the input value of the initial state unit at time point b;

[0118] Step S324: Design update unit, which is expressed as follows:

[0119] ;

[0120] in, represents the output value of the update unit at time point b, and Represent the weight and bias of the update unit respectively, represents the input value of the update unit at time point b;

[0121] Step S325: Design the output unit, as shown below:

[0122] ;

[0123] in, represents the output value of the output unit at time b, and denote the weight and bias of the output unit respectively, Represents the input value of the output unit at time point b;

[0124] Step S326: Output the state factor, which is expressed as follows:

[0125] ;

[0126] in, Represents the state factor of the memory layer at time b;

[0127] Step S33: Design a loss function, which is expressed as follows:

[0128] ;

[0129] in, 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, Indicates taking the absolute value;

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

[0131] By performing the above operations, the traditional dynamic prediction model of new energy electricity price 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 electricity price data. This scheme can more flexibly handle the nonlinear characteristics of new energy electricity price data by designing an output control function, and can capture the time series dependence and dynamic change characteristics in new energy electricity price data by designing a memory layer structure, which effectively solves the problems existing in traditional models when processing time series data and improves the prediction accuracy and generalization ability of the model.

[0132] Example 4, see Figure 1 This embodiment is based on the above embodiment. The price management is achieved by collecting new energy power generation data, meteorological data, grid status data and cost data, inputting the data into a dynamic prediction model for new energy electricity prices, and the model predicts the electricity price, and adjusting the electricity price predicted by the model to the real-time electricity price.

[0133] Example 5, see Figure 1 and Figure 2, this embodiment is based on the above embodiment, the artificial intelligence-based new energy power price management method and system provided by the present invention includes a data acquisition module, a data expansion module, a module for building a dynamic prediction model for new energy power prices and a price management module;

[0134] The data acquisition module collects historical renewable energy power generation data, meteorological data, grid status data, cost data and electricity 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 module for constructing a dynamic prediction model for new energy power prices;

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

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

[0138] It should be noted that, in this article, relational terms such as first and second, etc. 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 these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0139] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0140] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A new energy power price management method based on artificial intelligence, characterized by: The method comprises the following steps: Step S1: Data collection, collecting historical new energy power generation data, meteorological data, grid status data, cost data and electricity prices; the new energy power generation data includes power generation and power generation; the grid status 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; Among them, building a generator specifically includes the following steps: Step S211: Generate an embedding vector, which is expressed as follows: ； in, represents the embedding vector, represents the linear rectification function, and denote the noise weight matrix and the time weight matrix respectively, represents random noise, v represents the time step, represents the embedding bias; Step S212: construct a temporal convolution module, which is expressed as follows: ； in, 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, represents the convolution padding term, Representation vector Passing step length and the padding item after the element is adjusted. Represents the jth element of the convolution kernel; Step S213: construct a deconvolution module, which is expressed as follows: ； in, represents the output of the deconvolution module, j represents the index of the convolution kernel element, represents the deconvolution kernel size, represents the one-dimensional deconvolution step size, represents the deconvolution padding term, Representation vector After deconvolution step and deconvolution padding The adjusted elements, represents the deconvolution kernel elements; Step S3: construct a new energy power price dynamic prediction model by designing an output control function, a memory layer and a loss function; Step S4: Price management: Use the new energy power price dynamic prediction model to predict the power price, and adjust the power price predicted by the model to the real-time power price.

2. The artificial intelligence-based new energy power price management method according to claim 1 is characterized by: In step S2, the data expansion specifically includes the following steps: Step S21: construct a generator; Step S22: constructing a discriminator, specifically including the following steps: Step S221: construct a feature learning module, which is expressed as follows: ； in, represents the output of the feature learning module, represents the rectified linear function with leakage, u represents the layer index of the feature learning module, 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; Step S222: Design the attention layer, which is expressed as follows: ； in, represents the attention weight of the vth time step, represents the normalized exponential function, represents the 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-1th time step, represents the adjustment weight of the v-1th time step, represents the bias of the attention layer; Step S23: construct a loss function, which is expressed as follows: ； in, represents the loss of the generator, y represents the input of the generator, represents the probability distribution of the generator input, represents a generator, represents the discriminator, represents taking the logarithm, It means that when y is distributed from When randomly selecting from , take the expected value in brackets. 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 distribution of the real data, represents the loss of the discriminator, Represents real data, represents the probability distribution of real data, Indicates when From the distribution When randomly selecting from , take the expected value in brackets; Step S24: Data expansion, by using a generator to generate new energy power generation data, meteorological data, grid status data, cost data and electricity price data to expand the original data.

3. The artificial intelligence-based new energy power price management method according to claim 1 is characterized in that: In step S3, the construction of a new energy power price dynamic prediction model specifically includes the following steps: Step S31: Design an output control function, which is expressed as follows: ； in, represents the input value of the output control function, Represents the output value of the output control function, represents the sigmoid function, represents the logarithmic function with a natural constant as the base, represents the inverse tangent function, Indicates square root; Step S32: Designing a memory layer, specifically including the following steps: Step S321: Design the input unit, as shown below: ； Among them, b represents the index of the time point of the memory layer, represents the output value of the input unit at time point b, and denote the weight and bias of the input unit respectively, represents the state factor of the memory layer at time b-1, represents the input value of the input unit at time point b; Step S322: Design a forgetting unit, which is expressed as follows: ； in, represents the output value of the forgetting unit at time point b, and denote the weight and bias of the forgetting unit respectively, represents the input value of the forgetting unit at time point b; Step S323: construct an initial state unit, which is expressed as follows: ； in, represents the initial state factor at time point b, and They represent the initial state weight and initial state bias respectively, Represents the input value of the initial state unit at time point b; Step S324: Design update unit, which is expressed as follows: ； in, represents the output value of the update unit at time point b, and Represent the weight and bias of the update unit respectively, represents the input value of the update unit at time point b; Step S325: Design the output unit, as shown below: ； in, represents the output value of the output unit at time b, and denote the weight and bias of the output unit respectively, Represents the input value of the output unit at time point b; Step S326: Output the state factor, which is expressed as follows: ； in, Represents the state factor of the memory layer at time b; Step S33: Design a loss function, which is expressed as follows: ； in, 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, Indicates taking the absolute value; Step S34: Setting the model tag, and setting the electricity price as the tag data of the model.

4. The artificial intelligence-based new energy power price management method according to claim 1 is characterized in that: In step S1, the data collection is to collect historical new energy power generation data, meteorological data, grid status data, cost data and electricity prices; the new energy power generation data includes power generation and power generation; the grid status data includes normal state and abnormal state; the cost data includes construction cost and maintenance cost; In step S4, the price management is carried out by collecting new energy power generation data, meteorological data, grid status data and cost data, inputting the data into a dynamic prediction model for new energy power prices, and the model predicts the power price, and adjusting the power price predicted by the model to the real-time power price.

5. A new energy power price management system based on artificial intelligence, used to implement the new energy power price management method based on artificial intelligence as described in any one of claims 1 to 4, characterized in that: It includes data collection module, data expansion module, module for building dynamic prediction model of new energy power price and price management module; The data acquisition module collects historical renewable energy power generation data, meteorological data, grid status data, cost data and electricity prices, and sends the data to the data expansion module; 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 module for constructing a dynamic prediction model for new energy power prices; The module for constructing a dynamic prediction model for the price of new energy power receives data sent by the data expansion module, constructs a dynamic prediction model for the price of new energy power, and sends the data to the price management module; The price management module receives data sent by the module for constructing a dynamic prediction model for new energy power prices, predicts power prices using the dynamic prediction model for new energy power prices, and adjusts the power prices predicted by the model to real-time power prices.

Images