CNN-GRU neural network-based photovoltaic power plant carbon emission reduction prediction method and system

Abstract

The invention discloses a photovoltaic power plant carbon emission reduction prediction method and system based on a CNN-GRU neural network, and relates to the technical field of photovoltaic carbon emission reduction prediction, and the method comprises the steps: obtaining the original historical data of a photovoltaic power plant, carrying out the cleaning preprocessing of the original historical data of the photovoltaic power plant, and obtaining the original historical data of the photovoltaic power plant; performing normalization processing on the preprocessed original historical data of the photovoltaic power plant, and determining a characteristic value related to power output of the generating capacity by using a Pearson correlation coefficient to obtain a data set of the photovoltaic power plant; inputting the photovoltaic power plant data set into a pre-established CNN-GRU prediction network model for training, and obtaining a trained CNN-GRU prediction network model by adjusting hyper-parameters of the CNN-GRU prediction network model; and predicting a photovoltaic power plant data set by using the trained CNN-GRU prediction network model to obtain predicted photovoltaic power generation data, and performing carbon emission reduction prediction on the predicted photovoltaic power generation data by using a carbon emission reduction accounting methodology so as to effectively predict carbon emission reduction in the future.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic carbon emission reduction prediction, specifically to a method and system for predicting the carbon emission reduction of a photovoltaic power plant based on a CNN-GRU neural network. Background Art

[0002] The impact of global climate change on human society is increasing. As the global temperature continues to rise, extreme weather events occur frequently. In order to mitigate the impact of climate change, governments and enterprises around the world are actively promoting clean energy, such as solar energy, wind energy, etc. Among them, photovoltaic power generation, as a renewable clean energy, has become one of the research hotspots. Chinese Certified Emission Reduction (CCER) refers to the quantification and verification of the greenhouse gas emission reduction effects of projects such as renewable energy, forestry carbon sinks, and methane utilization in China, and the greenhouse gas emission reduction amounts registered in the national greenhouse gas voluntary emission reduction trading registration and registration system. With the development of the carbon market and the restart of China's national certified voluntary emission reduction mechanism, photovoltaic power generation can be used as an important means and tool for carbon emission reduction, and the carbon reduction amount can be used for the trading of national certified voluntary emission reduction amounts. However, since the power generation of photovoltaic power generation is greatly affected by weather and has certain volatility. Therefore, accurately predicting the carbon emission reduction amount generated by photovoltaic power generation and then using it for carbon market trading in the future is of great significance.

[0003] At present, many researchers have proposed different methods for predicting the carbon emission reduction of photovoltaic power generation, including time series methods and machine learning methods, etc. However, these methods have certain limitations in processing the feature extraction of meteorological data and the processing of time series data, and the prediction accuracy needs to be further improved. Summary of the Invention

[0004] To solve the deficiencies mentioned in the above background art, the purpose of the present invention is to provide a method and system for predicting the carbon emission reduction of a photovoltaic power plant based on a CNN-GRU neural network.

[0005] In the first aspect, the purpose of the present invention can be achieved through the following technical solutions: A method for predicting the carbon emission reduction of a photovoltaic power plant based on a CNN-GRU neural network, the method comprising the following steps:

[0006] Obtain the original historical data of the photovoltaic power plant, clean and preprocess the original historical data of the photovoltaic power plant, normalize the preprocessed original historical data of the photovoltaic power plant, and use the Pearson correlation coefficient to determine the eigenvalue related to the power output of the power generation amount to obtain the photovoltaic power plant data set;

[0007] Input the photovoltaic power plant dataset into the pre-established CNN-GRU prediction network model for training. By adjusting the hyperparameters of the CNN-GRU prediction network model, the trained CNN-GRU prediction network model is obtained;

[0008] Use the trained CNN-GRU prediction network model to predict the photovoltaic power plant dataset, obtain the predicted photovoltaic power generation data, and use the carbon emission reduction accounting methodology to predict the carbon emission reduction amount of the predicted photovoltaic power generation data.

[0009] Combined with the first aspect, in some implementation manners of the first aspect, the method further includes: the process of cleaning and preprocessing the original historical data of the photovoltaic power plant includes:

[0010] Check the duplicate abnormal data in the original historical data of the photovoltaic power plant, and then use the 3sigma criterion to eliminate the outliers in the original historical data of the photovoltaic power plant. The 3sigma criterion is based on the normal distribution and considers the data exceeding 3sigma as abnormal points. Use the 3sigma criterion to eliminate the outliers for each feature and the photovoltaic power generation value to obtain the preprocessed original historical data of the photovoltaic power plant.

[0011] Combined with the first aspect, in some implementation manners of the first aspect, the method further includes: the process of normalizing the preprocessed original historical data of the photovoltaic power plant and using the Pearson correlation coefficient to determine the eigenvalue related to the power output of the power generation amount to obtain the photovoltaic power plant dataset:

[0012] Establish a Pearson correlation coefficient table, analyze the relationship between different eigenvalues and the power generation power, explain the relationship between the preprocessed original historical data of the photovoltaic power plant, determine the input parameters with strong correlation with the photovoltaic power generation amount, and combine the photovoltaic power generation input features to select the input eigenvalues to obtain the photovoltaic power plant dataset. Divide the photovoltaic power plant dataset into a training set and a test set according to a preset ratio.

[0013] Combined with the first aspect, in some implementation manners of the first aspect, the method further includes: the pre-established CNN-GRU prediction network model includes a convolutional neural network CNN and a gated recurrent unit GRU. The CNN is mainly composed of a convolutional layer, a pooling layer, and a fully connected layer, and uses the methods of local perception and weight sharing to effectively extract the feature quantities of the original data. The structure of the GRU consists of two parts: a reset gate and an update gate. The reset gate is used to control the influence of the information at the previous moment on the current moment, and the update gate is used to control the influence of the information at the previous moment and the input at the current moment on the current moment.

[0014] In combination with the first aspect, in certain implementations of the first aspect, the method further includes: The CNN convolution operation includes the following formula:

[0015]

[0016] where y i,j is a point in the features output by the convolutional layer, x i,j is a pixel point in the input data, w k is the weight of the convolutional kernel, b is the bias, σ is the activation function, and K is the size of the convolutional kernel;

[0017] The GRU gated recurrent unit includes the following formula:

[0018] r t = σ(W ir x t + W hr h t-1 + b r )

[0019] z t = σ(W iz x t + W hz h t-1 + b z )

[0020]

[0021] where r t is the reset gate, z t is the update gate, is the candidate hidden state at the current time, h t is the hidden state at the current time, x t is the input at the current time, h t-1 is the hidden state at the previous time, W and b are the weights and biases, ⊙ is the element-wise multiplication operation, σ is the sigmoid function, and tanh is the hyperbolic tangent function.

[0022] In combination with the first aspect, in certain implementations of the first aspect, the method further includes: The process of the carbon emission reduction accounting methodology includes:

[0023] Determine the boundary of the project, that is, confirm the power generation scope, time, and space of the project, then identify the baseline of the project, compare the power generation of the project with the power generation under the preset power generation method, after determining the baseline, conduct an additionality demonstration: prove that the photovoltaic power generation of the project is based on newly built facilities, or the renovation or update of existing facilities, calculate the emission reduction of the project, and calculate the greenhouse gas emissions that the project can reduce by comparing the difference between the project and the baseline.

[0024] In combination with the first aspect, in certain implementations of the first aspect, the method further includes: the evaluation metric of the trained CNN-GRU prediction network model is the root mean square error, and the formula for the root mean square error is as follows:

[0025]

[0026] where n is the number of samples, y i is the true value, is the predicted value, and the calculation process of RMSE is to first calculate the prediction error of each sample, and then take the square root of the average of the prediction errors of all samples.

[0027] In a second aspect, to achieve the above object, the present invention discloses a photovoltaic power plant carbon emission reduction prediction system based on a CNN-GRU neural network, including:

[0028] A data processing module, configured to obtain the original historical data of the photovoltaic power plant, perform cleaning and preprocessing on the original historical data of the photovoltaic power plant, perform normalization processing on the preprocessed original historical data of the photovoltaic power plant, and use the Pearson correlation coefficient to determine the eigenvalue related to the power generation power output, so as to obtain a photovoltaic power plant dataset;

[0029] A model training module, configured to input the photovoltaic power plant dataset into a pre-established CNN-GRU prediction network model for training, and obtain a trained CNN-GRU prediction network model by adjusting the hyperparameters of the CNN-GRU prediction network model;

[0030] A carbon emission reduction prediction module, configured to use the trained CNN-GRU prediction network model to predict the photovoltaic power plant dataset, obtain the predicted photovoltaic power generation data, and use the carbon emission reduction accounting methodology to predict the carbon emission reduction amount of the predicted photovoltaic power generation data.

[0031] In another aspect of the present invention, to achieve the above object, a terminal device is disclosed, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The computer program stored in the memory is capable of running on the processor. When the processor loads and executes the computer program, the above-mentioned photovoltaic power plant carbon emission reduction prediction method based on a CNN-GRU neural network is adopted.

[0032] In still another aspect of the present invention, to achieve the above object, a computer-readable storage medium is disclosed. The computer-readable storage medium stores a computer program. When the computer program is loaded and executed by a processor, the above-mentioned photovoltaic power plant carbon emission reduction prediction method based on a CNN-GRU neural network is adopted.

[0033] Advantages of the present invention:

[0034] The present invention uses a convolutional neural network for feature extraction, which can better capture the features of meteorological data and improve the prediction accuracy. Secondly, the gated recurrent unit neural network can better process time series data and further improve the prediction accuracy. Finally, this method can be applied to the prediction of carbon emission reduction in other fields, such as wind power generation, etc. It can effectively predict future carbon emission reduction and provide support for optimizing energy production. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings;

[0036] Figure 1 is a schematic diagram of the method flow of the present invention;

[0037] Figure 2 is a schematic diagram of the working process of the present invention;

[0038] Figure 3 is a schematic diagram of the CNN-GRU algorithm model structure of the present invention;

[0039] Figure 4 is a diagram of the GRU gated recurrent unit of the present invention;

[0040] Figure 5 is a comparison diagram of the actual output of photovoltaic power and the predicted value of the CNN-GRU algorithm in the embodiment of the present invention;

[0041] Figure 6 is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a 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 those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0043] Embodiment 1:

[0044] As Figure 1 shown, a method for predicting carbon emission reduction of a photovoltaic power plant based on a CNN-GRU neural network, the method includes the following steps:

[0045] S101: acquiring original historical data of a photovoltaic power plant, cleaning and preprocessing the original historical data of the photovoltaic power plant, normalizing the preprocessed original historical data of the photovoltaic power plant, and using the Pearson correlation coefficient to determine characteristic values related to power output of power generation, thereby obtaining a data set of the photovoltaic power plant;

[0046] Specifically, the scheme of the present invention is further described below through an embodiment: In this embodiment, the photovoltaic power generation data of a photovoltaic power plant for 2 years is taken as the original photovoltaic power generation data. The parameters of the photovoltaic panels include: photovoltaic output power; weather parameters include: irradiance, temperature, wind speed, wind direction, and air pressure. The sampling time interval of these data is 15 minutes / time, which is collected by the monitoring system or data acquisition system of the photovoltaic power station, and stored, processed and analyzed through the cloud platform or other data management system.

[0047] In this embodiment, the 3sigma principle is used to remove outliers in the data. This principle is based on normal distribution and considers data exceeding 3sigma as outliers. The 3sigma principle is used to remove outliers for each feature and photovoltaic power value.

[0048] As shown in Table 1, a Pearson correlation coefficient table is established to analyze the relationship between different eigenvalues and power generation, understand and explain the relationship between the data, calculate the Pearson correlation coefficient between different eigenvalues and power generation, and organize it into a table to analyze and explain the relationship between the data. Among them, the input parameters with strong correlation between irradiance and active power have a correlation coefficient of 0.922; the correlation coefficient between wind speed and active power is 0.290; the correlation coefficient between temperature and active power is 0.291; the correlation coefficient between air pressure and active power is 0.025; wind direction and active power are negatively correlated, and the correlation coefficient is -0.135. Then, the processed data is divided into training set and test set according to the proportion. The training set is used to train the model, and the test set is used to evaluate the performance of the model.

[0049] Table 1 Pearson correlation coefficients of relevant features

[0050] Active power 0.29 0.291 0.025 0.922 -0.135 1 Wind direction 0.095 -0.003 -0.155 -0.089 1 -0.135 Irradiance 0.28 0.322 0.036 1 -0.089 0.922 Air pressure 0.049 -0.71 1 0.036 -0.155 0.025 Temperature 0.093 1 -0.71 0.322 -0.003 0.291 Wind speed 1 0.093 0.049 0.28 0.095 0.29 Wind speed Temperature Air pressure Irradiance Wind direction Active power

[0051] S102: inputting the photovoltaic power plant data set into a pre-established CNN-GRU prediction network model for training, and obtaining a trained CNN-GRU prediction network model by adjusting the hyperparameters of the CNN-GRU prediction network model;

[0052] like Figure 3As shown, the pre-established CNN-GRU prediction network model contains a convolutional neural network (CNN) and a gated recurrent unit (GRU). This model consists of a CNN network and a GRU network. This neural network is composed of a convolutional layer and a pooling layer. The convolution method selects SAME convolution, and the activation function is ELU. After convolution, the output feature map is input into a recurrent neural network with a single-layer GRU structure for learning to extract feature vectors. The output of the final fully connected layer obtains the final prediction value through inverse normalization. During the training process of the GRU recurrent neural network, the ADAM algorithm is used to iteratively update the weights, and the weights and biases of each neuron are continuously adjusted through momentum and adaptive learning rate to minimize the output value of the loss function and achieve the optimal effect. The present invention uses RMSE as the loss function to evaluate the model training situation. When training the model, appropriate hyperparameters such as learning rate, batch size, and number of iterations are selected, and cross-validation is performed to evaluate the model performance, and the trained model is saved for subsequent use.

[0053] The CNN convolution operation includes the following formula:

[0054]

[0055] Where, y i,j is a point in the features output by the convolutional layer, x i,j is a pixel point in the input data, w k is the weight of the convolution kernel, b is the bias, σ is the activation function, and K is the size of the convolution kernel.

[0056] The GRU gated recurrent unit is as Figure 4 shown and includes the following formula:

[0057] r t = σ(W ir x t + W hr h t-1 + b r )

[0058] z t = σ(W iz x t + W hz h t-1 + b z )

[0059]

[0060] Where, r t is the reset gate, z t is the update gate, is the candidate hidden state at the current moment, ht is the hidden state at the current moment, x t is the input at the current moment, h t-1 is the hidden state at the previous moment, W and b are weights and biases, ⊙ is the element-wise multiplication operation, σ is the sigmoid function, and tanh is the hyperbolic tangent function.

[0061] S103: Use the trained CNN-GRU prediction network model to predict the photovoltaic power plant dataset, obtain the predicted photovoltaic power generation data, and use the carbon emission reduction accounting methodology to predict the carbon emission reduction amount of the predicted photovoltaic power generation data.

[0062] This example uses one year of photovoltaic data for model training. The test set is the photovoltaic power generation data for the next day. The prediction results are as Figure 5 shown. According to the prediction results, the root mean square error of the test set is 4.2762, and the R-squared value is 0.9554.

[0063] This carbon emission reduction accounting adopts the "CM-001-V01 Consolidated Baseline Methodology for Renewable Energy Generation Grid-Connected Projects", which includes multiple steps to determine the carbon emission reduction amount of the photovoltaic power generation project.

[0064] Project boundary:

[0065] The boundary of this photovoltaic power generation project includes the photovoltaic power generation facilities and their ancillary facilities, grid interface facilities, and land and water resources related to the power generation facilities. Specifically, this photovoltaic power generation project is located in a rural area of Ningxia, China, covering an area of about 100 mu, with a total installed capacity of 10 MW. The grid interface of this project is the 110 kV transmission line in this area.

[0066] Project baseline:

[0067] The baseline of this photovoltaic power generation project refers to the power supply and demand situation and grid operation situation in this area without this project. According to the provisions of the first edition of the "Renewable Energy Grid-Connected Generation Methodology", the baseline of this project is the power generation capacity required in this area under the existing power supply and demand situation. After investigation and data analysis, the baseline of this project is 10 MW.

[0068] Additionality demonstration:

[0069] This photovoltaic power generation project meets the requirements of additionality demonstration. First, this project is constructed without other renewable energy projects, and there is no situation of being replaced by other projects. Second, the power generation capacity of this project exceeds the baseline of this area, with significant emission reduction benefits. Finally, the construction of this project requires a large amount of capital and technical support, with certain technological advancement and economic feasibility.

[0070] Carbon emission reduction calculation:

[0071] The carbon emission reduction amount of this photovoltaic power generation project is calculated using the method in "CM-001-V01 Integration Baseline for Renewable Energy Power Generation Grid-Connected Projects". Specifically, based on parameters such as the installed capacity, power generation, and electricity intensity of the project, the annual emission reduction amount of the project is calculated. After calculation, the annual emission reduction amount of this project is 5,000 tons of carbon dioxide.

[0072] Example 2: To achieve the above object, as Figure 6 shown, the present invention discloses a carbon emission reduction amount prediction system for a photovoltaic power plant based on a CNN-GRU neural network, including:

[0073] A data processing module 11, configured to obtain the original historical data of the photovoltaic power plant, perform cleaning and preprocessing on the original historical data of the photovoltaic power plant, perform normalization processing on the preprocessed original historical data of the photovoltaic power plant, and use the Pearson correlation coefficient to determine the eigenvalue related to the power output of the power generation, so as to obtain the photovoltaic power plant data set;

[0074] A model training module 12, configured to input the photovoltaic power plant data set into a pre-established CNN-GRU prediction network model for training, and obtain a trained CNN-GRU prediction network model by adjusting the hyperparameters of the CNN-GRU prediction network model;

[0075] A carbon emission reduction amount prediction module 13, configured to use the trained CNN-GRU prediction network model to predict the photovoltaic power plant data set, obtain the predicted photovoltaic power generation data, and use the carbon emission reduction accounting methodology to predict the carbon emission reduction amount of the predicted photovoltaic power generation data.

[0076] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is configured to execute the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and is used to implement one or more instructions. Specifically, it is used to load and execute one or more instructions in the computer storage medium to implement the above method.

[0077] It should be further noted that, based on the same inventive concept, the present invention also provides a computer storage medium, on which a computer program is stored, and when the computer program is run by a processor, it executes the above method. The storage medium may be any combination of one or more computer-readable media. The computer-readable media may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electro-magnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or combined with an instruction execution system, apparatus, or device.

[0078] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0079] The above shows and describes the basic principles, main features, and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited by the above embodiments. The above embodiments and the descriptions in the specification only illustrate the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure will have various changes and improvements, and these changes and improvements all fall within the scope of the present disclosure claimed.

Claims

1. A method for predicting the carbon emission reduction of a photovoltaic power plant based on a CNN-GRU neural network, characterized in that The method comprises the following steps: Acquire the original historical data of the photovoltaic power plant, clean and preprocess the original historical data of the photovoltaic power plant, normalize the preprocessed original historical data of the photovoltaic power plant, and use the Pearson correlation coefficient to determine the characteristic value related to the power output of the power generation, so as to obtain the photovoltaic power plant data set; The photovoltaic power plant data set is input into the pre-established CNN-GRU prediction network model for training, and the trained CNN-GRU prediction network model is obtained by adjusting the hyperparameters of the CNN-GRU prediction network model; The trained CNN-GRU prediction network model is used to predict the photovoltaic power plant data set to obtain the predicted photovoltaic power generation data, and the carbon emission reduction accounting methodology is used to predict the carbon emission reduction of the predicted photovoltaic power generation data.

2. The method for predicting the carbon emission reduction amount of a photovoltaic power plant based on a CNN-GRU neural network according to claim 1, wherein The process of cleaning and preprocessing the original historical data of the photovoltaic power plant includes: The repeated abnormal data in the original historical data of the photovoltaic power plant are checked, and then the 3sigma criterion is used to eliminate the abnormal values in the original historical data of the photovoltaic power plant. The 3sigma criterion is based on the normal distribution and considers that the data exceeding 3sigma is an abnormal point. The 3sigma criterion is used to eliminate the abnormal values of each feature and photovoltaic power generation value to obtain the preprocessed original historical data of the photovoltaic power plant.

3. The photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network according to claim 1, characterized in that The process of normalizing the pre-processed original historical data of the photovoltaic power plant and using the Pearson correlation coefficient to determine the characteristic values related to the power output of the power generation to obtain the photovoltaic power plant data set: A Pearson correlation coefficient table is established to analyze the relationship between different eigenvalues and power generation, explain the relationship between the preprocessed original historical data of the photovoltaic power plant, determine the input parameters with a strong correlation with the photovoltaic power generation, and select the input eigenvalues in combination with the photovoltaic power generation input characteristics to obtain the photovoltaic power plant data set, which is then divided into a training set and a test set according to a preset ratio.

4. The photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network according to claim 1, wherein, The pre-established CNN-GRU prediction network model includes a convolutional neural network CNN and a gated recurrent unit GRU. The CNN is mainly composed of a convolutional layer, a pooling layer and a fully connected layer. It adopts local perception and weight sharing to effectively extract the feature quantity of the original data. The structure of the GRU consists of a reset gate Reset Gate and an update gate Update Gate. The reset gate is used to control the influence of the information of the previous moment on the current moment, and the update gate is used to control the influence of the information of the previous moment and the input of the current moment on the current moment.

5. The photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network according to claim 4, characterized in that The CNN convolution operation includes the following formula: Among them, y i,j is a point in the features output by the convolutional layer, x i,j is a pixel point in the input data, w k is the weight of the convolutional kernel, b is the bias, σ is the activation function, and K is the size of the convolutional kernel; The GRU gated recurrent unit includes the following formula: r t = σ(W ir x t + W hr h t-1 + b r ) z t = σ(W iz x t + W hz h t-1 + b z ) where, r t is the reset gate, z t is the update gate, is the candidate hidden state at the current time step, h t is the hidden state at the current time step, x t is the input at the current time step, h t-1 is the hidden state at the previous time step, W and b are the weights and biases, ⊙ is the element-wise multiplication operation, σ is the sigmoid function, and tanh is the hyperbolic tangent function.

6. The photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network according to claim 1, wherein The process of the carbon reduction accounting methodology includes: Determine the boundaries of the project, that is, confirm the power generation scope, time, and space of the project, then identify the baseline of the project, compare the power generation of the project with the power generation under the preset power generation method, and after determining the baseline, conduct additionality demonstration: prove that the photovoltaic power generation of the project is based on new facilities, or the transformation or renewal of existing facilities, calculate the emission reduction of the project, and calculate the greenhouse gas emissions that the project can reduce by comparing the difference between the project and the baseline.

7. The photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network according to claim 1, wherein The evaluation index of the trained CNN-GRU prediction network model is the root mean square error, and the formula for the root mean square error is as follows: where n is the number of samples, and y i is the true value, is the predicted value. The calculation process of RMSE is to first calculate the prediction error of each sample, and then take the square root of the average value of the prediction errors of all samples.

8. A photovoltaic power plant carbon emission reduction prediction system based on a CNN-GRU neural network, characterized in that, Including: A data processing module, which is used to obtain the original historical data of the photovoltaic power plant, clean and preprocess the original historical data of the photovoltaic power plant, normalize the preprocessed original historical data of the photovoltaic power plant, and use the Pearson correlation coefficient to determine the eigenvalue related to the power output of the power generation, so as to obtain the photovoltaic power plant data set; A model training module, which is used to input the photovoltaic power plant data set into a pre-established CNN-GRU prediction network model for training, and obtain a trained CNN-GRU prediction network model by adjusting the hyperparameters of the CNN-GRU prediction network model; A carbon emission reduction prediction module, which is used to use the trained CNN-GRU prediction network model to predict the photovoltaic power plant data set, obtain the predicted photovoltaic power generation data, and use the carbon emission reduction accounting methodology to predict the carbon emission reduction of the predicted photovoltaic power generation data.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, it adopts the photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program therein, characterized in that, When the computer program is loaded and executed by the processor, it adopts the photovoltaic power plant carbon emission reduction prediction method based on the CNN-GRU neural network described in any one of claims 1 to 7.

Images