Power generation capacity pre-judgment method based on cloud computing integration technology

Abstract

The invention provides a generating capacity pre-judgment method based on a cloud computing integration technology, and relates to the field of power generation, and the method comprises the following steps: S100, collecting global climate model prediction data, power grid load data and power plant equipment health degree monitoring data in real time through a cloud computing platform; the global climate model prediction data comprises temperature, humidity, wind speed, wind direction and precipitation. The climate-power grid-equipment ternary fusion generating capacity pre-judgment system combining Al deep learning and cloud computing is an innovative solution, not only can analyze the relationship between historical meteorological data and generating capacity, but also can integrate global climate model prediction, power grid load dynamics and power plant equipment health degree monitoring data in real time; by introducing a microclimate simulation technology, high-precision prediction is performed on local climate around a power plant, and power generation efficiency prediction is dynamically adjusted in combination with an equipment aging prediction model, so that accurate pre-judgment is realized.

Description

Technical Field

[0001] The present invention relates to the field of power generation, and in particular to a method for predicting power generation based on cloud computing integration technology. Background Art

[0002] With the rapid development of the energy industry and the advancement of smart grid technology, power generation forecasting has become an important part of power dispatching and resource management. Equipment aging will lead to a gradual decline in power generation efficiency, but this process is often gradual and difficult to predict. Climate change has a significant impact on power generation. The complexity and uncertainty of the climate itself, due to the aging of equipment and the impact of climate change, traditional power generation forecasting methods mainly rely on statistical analysis of historical data. Although this method is simple and easy, its prediction accuracy is often limited when faced with the complex and changeable modern power grid operating environment.

[0003] Therefore, we make improvements to this and propose a power generation prediction method based on cloud computing integration technology. Summary of the invention

[0004] The purpose of the present invention is to address the problem that the existing traditional power generation prediction method mainly relies on statistical analysis of historical data. Although this method is simple and easy to implement, its prediction accuracy is often limited when facing the complex and changeable modern power grid operating environment.

[0005] In order to achieve the above-mentioned purpose of the invention, the present invention provides a power generation prediction method based on cloud computing integration technology to improve the above-mentioned problem.

[0006] The specific application is as follows:

[0007] The following steps are involved:

[0008] S100, collects global climate model prediction data, power grid load data and power plant equipment health monitoring data in real time through cloud computing platform;

[0009] Global climate model prediction data include temperature, humidity, wind speed, wind direction, and precipitation;

[0010] Grid load data includes real-time power consumption, peak power consumption periods, and valley power consumption periods;

[0011] Power plant equipment health monitoring data includes equipment operating time, maintenance records, failure frequency, and performance indicators;

[0012] Use microclimate simulation technology to predict the climate in the area where the power plant is located;

[0013] S200, preprocessing the collected data, including data cleaning, standardization and formatting;

[0014] S300, combining the equipment's operating time, maintenance records, failure frequency, and performance index data, using machine learning algorithms to establish an equipment aging prediction model;

[0015] S400, using Al deep learning algorithm to perform data analysis, integrating the microclimate simulation results and the output results of the equipment aging prediction model, to obtain the power generation efficiency prediction value.

[0016] As the preferred technical solution of this application, the microclimate simulation technology is used to predict the climate in the area where the power plant is located, which specifically includes the following steps:

[0017] S101, determine model parameters:

[0018] The atmospheric pressure is P;

[0019] The temperature is T;

[0020] Humidity is q;

[0021] The wind speed in the east-west direction is u, and the wind speed in the north-south direction is v;

[0022] The precipitation is R.

[0023] As the preferred technical solution of this application, the use of microclimate simulation technology to predict the climate in the area where the power plant is located also includes the following steps:

[0024] S102. Constructing a set of physical equations:

[0025] Continuity equation:

[0026] Where ρ is the air density and u is the wind speed vector;

[0027] Momentum equation:

[0028] Where P is the pressure, g is the acceleration due to gravity, and τ is the stress tensor;

[0029] Energy equation:

[0030] Where e is the total energy, k is the thermal conductivity, and Q represents other heat source terms (additional heat energy sources other than physical processes);

[0031] Water vapor equation:

[0032] Among them, E is the evaporation rate and C is the condensation rate;

[0033] S103, numerical solution:

[0034] Solve the partial differential equations in S102, advance time step by step through the numerical method, calculate the numerical solutions of various physical quantities in the atmosphere at different time and space positions, and update the variable X (parameter in S101) at the grid point (i, j) as follows:

[0035]

[0036] Among them, F is the flux function (describing the influence of the surrounding grid points on the central point and reflecting the propagation characteristics of the physical quantity X), D is the diffusion term (describing the natural diffusion characteristics of the physical quantity X), and Δt is the time step.

[0037] As the preferred technical solution of this application, the use of microclimate simulation technology to predict the climate in the area where the power plant is located also includes the following steps:

[0038] S104. Model calibration and validation:

[0039] Calibrate and validate the model by comparing the model output with actual observed data, and use statistical methods to assess the accuracy of the prediction results;

[0040] S105, prediction result output:

[0041] Finally, the calculated results are presented in the form of graphs, including temperature distribution graph, wind field graph, and precipitation distribution graph.

[0042] As the preferred technical solution of this application, a machine learning algorithm is used to establish an equipment aging prediction model, which specifically includes:

[0043] S301. Data collection:

[0044] Collect grid load data and power plant equipment health monitoring data;

[0045] S302, model training:

[0046] The dataset is divided into training set and test set with a ratio of 80% and 20%;

[0047] S303, Model Evaluation:

[0048] The performance of the model was evaluated using mean square error and coefficient of determination;

[0049] S304, Model Application:

[0050] Use the trained model to predict the aging of future equipment.

[0051] As a preferred technical solution of the present application, S302 specifically includes the following training steps:

[0052] S3020. Define the loss function:

[0053] For a given data point (x i ,y i ), the loss function is defined as:

[0054] L ε (y i -f(x i ))=max(0,|y i -f(x i )|-ε);

[0055] S3021. Construct objective function:

[0056] The goal is to minimize the sum of the losses of all samples plus a penalty for model complexity, expressed as: Among them, w and b are the weight vector and bias term respectively, and C is the regularization parameter used to balance the loss and regularization term. and are the upper and lower bound slack variables, representing the positive and negative errors of each sample beyond ε, w T is the transpose of the weight vector w, w T w is the square of the length (modulus) of the weight vector w, and n is the number of samples.

[0057] As a preferred technical solution of the present application, S302 specifically includes the following training steps:

[0058] S3022. Set constraints:

[0059]

[0060] For each sample point, the constraints are expressed as: Here, φ(x i ) is the mapping function from the original feature space to the higher dimensional feature space, (x i ) is the feature vector of the i-th sample. In practical applications, kernel techniques are used to avoid directly calculating this mapping. The kernel function k(x i , x j )=φ(x i ) T φ(x j ) to simplify the calculation.

[0061] As a preferred technical solution of the present application, S302 specifically includes the following training steps:

[0062] S3023, obtain the prediction function:

[0063] Through S3020~S3022, the final weight vector w and bias term b are obtained, and then the prediction function is obtained:

[0064] f(x)=w T φ(x)+b

[0065] As a preferred technical solution of this application, in S104, the evaluation is performed using the following formula:

[0066] Mean Squared Error (MSE): It measures the average squared difference between the predicted value and the true value, where n is the number of observations and y is i is the true value of the ith observation, is the model's predicted value for the i-th observation. The smaller the mean square error, the better the prediction effect of the model.

[0067] Coefficient of determination (R 2 score): It measures the percentage reduction in the difference between the model's predictions and the true values, where SSE is the residual sum of squares, that is, the sum of the squares of the differences between the predicted values ​​and the true values: SST is the total sum of squares, which is the sum of the squares of the differences between the true value and the mean true value: is the average of all true values. The range of the coefficient of determination is between [0, 1]. When the coefficient of determination is close to 1, it indicates that the model predicts very well. 2 When the score is close to 0, the model's predictive ability is weak, the coefficient of determination is less than 0, and the constant mean value is used as the prediction.

[0068] As a preferred technical solution of the present application, in S400, the method for predicting the power generation efficiency prediction value includes the following steps:

[0069] S401. Collect microclimate simulation results and equipment aging prediction results;

[0070] S402, constructing a prediction model:

[0071] The input layer receives the multimodal time series data of microclimate simulation results and equipment aging prediction results as input;

[0072] Convolutional layers are used to extract spatial features from input data, such as the spatial distribution pattern of meteorological data;

[0073] The pooling layer is used to reduce data dimensions, reduce computational effort, and improve model efficiency;

[0074] The LSTM layer is used to process time series data after being processed by the convolutional layer and capture long-term dependencies, such as how device performance changes over time;

[0075] The fully connected layer is used to integrate the information output by the LSTM layer and transform it into the final prediction output.

[0076] As a preferred technical solution of the present application, S400 further includes the following steps:

[0077] S403, compile model:

[0078] Select the Adam optimizer to compile the model to improve the stability and convergence speed of model training;

[0079] S404, power generation efficiency model training:

[0080] Use the training set data to train the model and monitor the performance of the model on the validation set to prevent overfitting;

[0081] S405, Power Generation Efficiency Model Evaluation:

[0082] Use the test set to evaluate the model performance and calculate the error between the predicted power generation efficiency and the actual power generation efficiency;

[0083] S406. Predicting power generation efficiency:

[0084] Use the trained model to predict the new time series data and obtain the predicted value of power generation efficiency.

[0085] As a preferred technical solution of the present application, the convolution layer is calculated. For each convolution kernel, the result of the convolution operation is calculated as:

[0086]

[0087] Among them, I represents the input image, K represents the convolution kernel, and M and N represent the sizes of the convolution kernel;

[0088] As the preferred technical solution of this application, the pooling layer is calculated, and for the maximum pooling operation, the output is:

[0089] Among them, P represents the output after pooling, I represents the input image, and M and N represent the size of the pooling window;

[0090] As the preferred technical solution of this application, the internal operation of the LSTM unit;

[0091] Forget gate calculation: f t =σ(W f ·[h t-1 , x t ]+b f );

[0092] Input gate calculation: i t =σ(W i ·[h t-1 , x t ]+b i);

[0093] Candidate cell state calculation:

[0094] Current cell state calculation:

[0095] Output gate calculation: o t =σ(W o ·[h t-1 , x t ]+b o );

[0096] Hidden state calculation; h t =o t *tanh(c t ).

[0097] As a preferred technical solution of the present application, the predicted power generation efficiency is expressed as P=M(X), where M is a trained model, X is a set of feature vectors, and P is the predicted power generation efficiency.

[0098] Compared with the prior art, the present invention has the following beneficial effects:

[0099] In the scheme of this application:

[0100] 1. The climate-grid-equipment ternary fusion power generation prediction system that combines AI deep learning and cloud computing is an innovative solution. It can not only analyze the relationship between historical meteorological data and power generation, but also integrate global climate model predictions, grid load dynamics, and power plant equipment health monitoring data in real time. By introducing microclimate simulation technology, it can make high-precision predictions of the local climate around the power plant. Combined with the equipment aging prediction model, it can dynamically adjust the power generation efficiency prediction to achieve accurate prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 A schematic diagram of a power generation prediction method based on cloud computing integration technology provided in this application. DETAILED DESCRIPTION

[0102] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0103] As described in the background art, the traditional power generation forecasting method mainly relies on the statistical analysis of historical data. Although this method is simple and easy to implement, its forecasting accuracy is often limited when facing the complex and changeable modern power grid operating environment.

[0104] In order to solve this technical problem, the present invention provides a method for predicting power generation based on cloud computing integration technology, which is applied to the field of power generation.

[0105] Specifically, please refer to Figure 1 The power generation prediction method based on cloud computing integration technology specifically includes the following steps:

[0106] S100, collects global climate model prediction data, power grid load data and power plant equipment health monitoring data in real time through cloud computing platform;

[0107] Global climate model prediction data include temperature, humidity, wind speed, wind direction, and precipitation;

[0108] Grid load data includes real-time power consumption, peak power consumption periods, and valley power consumption periods;

[0109] Power plant equipment health monitoring data includes equipment operating time, maintenance records, failure frequency, and performance indicators;

[0110] Use microclimate simulation technology to predict the climate in the area where the power plant is located;

[0111] S200, preprocessing the collected data, including data cleaning, standardization and formatting;

[0112] S300, combining the equipment's operating time, maintenance records, failure frequency, and performance index data, using machine learning algorithms to establish an equipment aging prediction model;

[0113] S400 uses AI deep learning algorithms to perform data analysis, integrates the microclimate simulation results and the output results of the equipment aging prediction model, and obtains the power generation efficiency prediction value.

[0114] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings.

[0115] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features and technical solutions in the embodiments may be combined with each other.

[0116] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0117] Embodiment 1, a method for predicting power generation based on cloud computing integration technology, using microclimate simulation technology to predict the climate in the area where the power plant is located, specifically includes the following steps:

[0118] S101, determine model parameters:

[0119] The atmospheric pressure is P;

[0120] The temperature is T;

[0121] Humidity is q;

[0122] The wind speed in the east-west direction is u, and the wind speed in the north-south direction is v;

[0123] The precipitation is R.

[0124] The use of microclimate simulation technology to predict the climate in the power plant area also includes the following steps:

[0125] S102. Constructing a set of physical equations:

[0126] Construct a set of partial differential equations to describe the variation of parameters in S101 with time and space;

[0127] Continuity equation:

[0128] Where ρ is the air density and u is the wind speed vector;

[0129] Momentum equation:

[0130] Where P is the pressure, g is the acceleration due to gravity, and τ is the stress tensor;

[0131] Energy equation:

[0132] Where e is the total energy, k is the thermal conductivity, and Q represents other heat source terms (additional heat energy sources other than physical processes);

[0133] Water vapor equation:

[0134] Among them, E is the evaporation rate and C is the condensation rate;

[0135] S103, numerical solution:

[0136] Solve the partial differential equations in S102, advance time step by step through the numerical method, calculate the numerical solutions of various physical quantities in the atmosphere at different time and space positions, and update the variable X (parameter in S101) at the grid point (i, j) as follows:

[0137]

[0138] Among them, F is the flux function (describing the influence of the surrounding grid points on the central point, reflecting the propagation characteristics of the physical quantity X), D is the diffusion term (describing the natural diffusion characteristics of the physical quantity X), and Δt is the time step. Through the above formula, the changes of various physical quantities in the atmosphere over time can be simulated, and the temperature changes, humidity changes, wind direction and speed changes within the future Δt time can be predicted;

[0139] S104. Model calibration and validation:

[0140] Calibrate and validate the model by comparing the model output with actual observed data, and use statistical methods to assess the accuracy of the prediction results;

[0141] S105, prediction result output:

[0142] Finally, the calculated results are presented in the form of graphs, including temperature distribution graph, wind field graph, and precipitation distribution graph.

[0143] Embodiment 2 further optimizes the power generation prediction method based on cloud computing integration technology provided in Embodiment 1. Specifically, a machine learning algorithm is used to establish an equipment aging prediction model, which specifically includes:

[0144] S301. Data collection:

[0145] Collect grid load data and power plant equipment health monitoring data;

[0146] S302, model training:

[0147] The dataset is divided into training set and test set with a ratio of 80% and 20%;

[0148] S303, Model Evaluation:

[0149] The mean square error (MSE), coefficient of determination (R 2 score) to evaluate the performance of the model;

[0150] S304, Model Application:

[0151] Use the trained model to predict the aging of future equipment.

[0152] S302 specifically includes the following training steps:

[0153] S3020. Define the loss function:

[0154] For a given data point (x i ,y i ), the loss function is defined as:

[0155] Lε (y i -f(x i ))=max(0,|y i -f(x i )|-ε)f(x i ) is the predicted value, ε is the error threshold, if the error is less than ε, the loss is 0; otherwise, the loss is the part exceeding ε;

[0156] S3021. Construct objective function:

[0157] The goal is to minimize the sum of the losses of all samples plus the penalty term (regularization term) of model complexity, expressed as: Among them, w and b are the weight vector and bias term respectively, and C is the regularization parameter used to balance the loss and regularization term. and are the upper and lower bound slack variables, representing the positive and negative errors of each sample beyond ε, w T is the transpose of the weight vector w, w T w is the square of the length (modulus) of the weight vector w, and n is the number of samples.

[0158] In S302, the following training steps are specifically included:

[0159] S3022. Set constraints:

[0160]

[0161] For each sample point, the constraints are expressed as: Here, φ(x i ) is the mapping function from the original feature space to the higher dimensional feature space, (x i ) is the feature vector of the i-th sample. In practical applications, kernel techniques are used to avoid directly calculating this mapping. The kernel function k(x i , x j )=φ(x i ) T φ(x j ) to simplify the calculation.

[0162] S3023. Select kernel function:

[0163] Select one of the linear kernel, polynomial kernel and Gaussian kernel;

[0164] The linear kernel is expressed as:

[0165] The polynomial kernel is expressed as:

[0166] The Gaussian kernel is expressed as: k(xi , x j ) = exp(-γ||x i -x j || 2 ), where γ>0 is the parameter of the Gaussian kernel, which determines the width of the Gaussian kernel, k(x i , x j ) is the kernel function, which measures the two samples x i and x j Similarity in high-dimensional space, d is the degree of the polynomial kernel function, and γ is the parameter in the Gaussian kernel function, which determines the width of the Gaussian kernel.

[0167] In S302, the following training steps are specifically included:

[0168] S3023, obtain the prediction function:

[0169] Through S3020~S3022, the final weight vector w and bias term b are obtained, and then the prediction function is obtained:

[0170] f(x)=w T φ(x)+b

[0171] If Gaussian kernel is used for regression prediction, the final prediction function can be written as:

[0172] in, are Lagrange multipliers, which can be obtained by solving the above optimization problem.

[0173] In Example 3, the power generation prediction method based on cloud computing integration technology provided in Example 1 or 2 is further optimized. In S104, the evaluation is performed using the following formula:

[0174] Mean Squared Error (MSE): It measures the average squared difference between the predicted value and the true value, where n is the number of observations and y is i is the true value of the ith observation, is the model's predicted value for the i-th observation. The smaller the mean square error, the better the prediction effect of the model.

[0175] Coefficient of determination (R 2 score): It measures the percentage reduction in the difference between the model's predictions and the true values, where SSE is the residual sum of squares, that is, the sum of the squares of the differences between the predicted values ​​and the true values: SST is the total sum of squares, which is the sum of the squares of the differences between the true value and the mean true value: is the average of all true values. The range of the coefficient of determination is between [0, 1]. When the coefficient of determination is close to 1, it indicates that the model predicts very well. 2 When the score is close to 0, the model's predictive ability is weak, the coefficient of determination is less than 0, and the constant mean value is used as the prediction.

[0176] In S400, the method for predicting the power generation efficiency prediction value includes the following steps:

[0177] S401. Collect microclimate simulation results and equipment aging prediction results;

[0178] S402, constructing a prediction model:

[0179] The input layer receives the multimodal time series data of microclimate simulation results and equipment aging prediction results as input;

[0180] Convolutional layers are used to extract spatial features from input data, such as the spatial distribution pattern of meteorological data;

[0181] The pooling layer is used to reduce data dimensions, reduce computational effort, and improve model efficiency;

[0182] The LSTM layer is used to process time series data after being processed by the convolutional layer and capture long-term dependencies, such as how device performance changes over time;

[0183] The fully connected layer is used to integrate the information output by the LSTM layer and transform it into the final prediction output.

[0184] S400 also includes the following steps:

[0185] S403, compile model:

[0186] Select the Adam optimizer to compile the model to improve the stability and convergence speed of model training;

[0187] S404, power generation efficiency model training:

[0188] Use the training set data to train the model and monitor the performance of the model on the validation set to prevent overfitting;

[0189] S405, Power Generation Efficiency Model Evaluation:

[0190] Use the test set to evaluate the model performance and calculate the error between the predicted power generation efficiency and the actual power generation efficiency;

[0191] S406. Predicting power generation efficiency:

[0192] Use the trained model to predict the new time series data and obtain the predicted value of power generation efficiency.

[0193] Convolution layer calculation, for each convolution kernel, the result of the convolution operation is calculated as:

[0194]

[0195] Among them, I represents the input image, K represents the convolution kernel, and M and N represent the sizes of the convolution kernel;

[0196] Pooling layer calculation, for the maximum pooling operation, the output is:

[0197] Among them, P represents the output after pooling, I represents the input image, and M and N represent the size of the pooling window;

[0198] Internal workings of LSTM cells;

[0199] Forget gate calculation: f t =σ(W f ·[h t-1 , x t ]+b f );

[0200] Input gate calculation: i t =σ(W i ·[h t-1 , x t ]+b i );

[0201] Candidate cell state calculation:

[0202] Current cell state calculation:

[0203] Output gate calculation: o t =σ(V o ·[h t-1 , x t ]+b o );

[0204] Hidden state calculation: h t =o t *tanh(c t ). Among them, W f , W i , W c and W o is the weight matrix, b f 、b i 、b c 、b o is the bias term, σ represents the sigmoi activation function, and tanh represents the hyperbolic tangent activation function.

[0205] The predicted power generation efficiency is expressed as P=M(X), where M is the trained model, X is a set of feature vectors, and P is the predicted power generation efficiency, where M is a deep learning model composed of a CNN layer and an LSTM layer, and X contains the feature vectors of time series data.

[0206] In the present invention, unless otherwise clearly defined and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral one; it can be a mechanical connection, an electrical connection, or communication with each other; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0207] Obviously, the embodiments described above are only some embodiments of the present invention, rather than all embodiments. The preferred embodiments of the present invention are given in the accompanying drawings, but they do not limit the patent scope of the present invention. The present invention can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive. Although the present invention has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific embodiments, or to perform equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of the present invention, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present invention.

Claims

1. A method for predicting power generation based on cloud computing integration technology, characterized in that: The following steps are involved: S100, collects global climate model prediction data, power grid load data and power plant equipment health monitoring data in real time through cloud computing platform; Global climate model prediction data include temperature, humidity, wind speed, wind direction, and precipitation; Grid load data includes real-time power consumption, peak power consumption periods, and valley power consumption periods; Power plant equipment health monitoring data includes equipment operating time, maintenance records, failure frequency, and performance indicators; Use microclimate simulation technology to predict the climate in the area where the power plant is located; S200, preprocessing the collected data, including data cleaning, standardization and formatting; S300, combining the equipment's operating time, maintenance records, failure frequency, and performance index data, using machine learning algorithms to establish an equipment aging prediction model; S400, using Al deep learning algorithm to perform data analysis, integrating the microclimate simulation results and the output results of the equipment aging prediction model, to obtain the power generation efficiency prediction value.

2. The power generation prediction method based on cloud computing integration technology according to claim 1 is characterized in that: The microclimate simulation technology is used to predict the climate in the area where the power plant is located, which includes the following steps: S101, determine model parameters: The atmospheric pressure is P; The temperature is T; Humidity is q; The wind speed in the east-west direction is u, and the wind speed in the north-south direction is v; The precipitation is R.

3. The power generation prediction method based on cloud computing integration technology according to claim 2 is characterized in that: The use of microclimate simulation technology to predict the climate in the power plant area also includes the following steps: S102. Constructing a set of physical equations: Continuity equation: Where ρ is the air density and u is the wind speed vector; Momentum equation: Where P is the pressure, g is the acceleration due to gravity, and τ is the stress tensor; Energy equation: Where e is the total energy, k is the thermal conductivity, and Q represents other heat source terms (additional heat energy sources other than physical processes); Water vapor equation: Among them, E is the evaporation rate and C is the condensation rate; S103, numerical solution: Solve the partial differential equations in S102, advance the time step by step through the numerical method, calculate the numerical solutions of various physical quantities in the atmosphere at different time and space positions, and update the variable X at the grid point (i, j) as follows:

4. The power generation prediction method based on cloud computing integration technology according to claim 3 is characterized in that: The use of microclimate simulation technology to predict the climate in the power plant area also includes the following steps: S104. Model calibration and validation: Calibrate and validate the model by comparing the model output with actual observed data, and use statistical methods to assess the accuracy of the prediction results; S105, prediction result output: Finally, the calculated results are presented in the form of graphs, including temperature distribution graph, wind field graph, and precipitation distribution graph.

5. The power generation prediction method based on cloud computing integration technology according to claim 1 is characterized in that: Use machine learning algorithms to build equipment aging prediction models, including: S301. Data collection: Collect grid load data and power plant equipment health monitoring data; S302, model training: Divide the dataset into training and testing sets; S303, Model Evaluation: The performance of the model was evaluated using mean square error and coefficient of determination; S304, Model Application: Use the trained model to predict the aging of future equipment.

6. The power generation prediction method based on cloud computing integration technology according to claim 5 is characterized in that: S302 specifically includes the following training steps: S3020. Define the loss function: For a given data point (xi, yi), the loss function is defined as: L ε (is f(x i ))=max(0,|f(x i )|−ε) S3021. Construct objective function: The goal is to minimize the sum of the losses of all samples plus a penalty for model complexity, expressed as:

7. The power generation prediction method based on cloud computing integration technology according to claim 6 is characterized in that: In S302, the following training steps are specifically included: S3022. Set constraints: For each sample point, the constraints are expressed as: S3023. Select kernel function: Select one of the linear kernel, polynomial kernel and Gaussian kernel; The linear kernel is expressed as: The polynomial kernel is expressed as: The Gaussian kernel is expressed as: k(x i , x j ) = exp(-γ||x i -x j || 2 ).

8. The power generation prediction method based on cloud computing integration technology according to claim 7 is characterized in that: In S302, the following training steps are specifically included: S3023, obtain the prediction function: Through S3020~S3022, the final weight vector w and bias term b are obtained, and then the prediction function is obtained: f(x)=w T φ(x)+b 9. The power generation prediction method based on cloud computing integration technology according to claim 4 is characterized in that: In S104, the evaluation is performed using the following formula: Mean Square Error: It measures the average squared difference between the predicted value and the true value, where n is the number of observations and y is i is the true value of the ith observation, is the model's prediction for the ith observation; Coefficient of determination (R 2 score): It measures the percentage reduction in the difference between the model's predictions and the true values, where SSE is the residual sum of squares, that is, the sum of the squares of the differences between the predicted values ​​and the true values: SST is the total sum of squares, which is the sum of the squares of the differences between the true value and the mean true value: It is the average of all true values, and the range of the coefficient of determination is between [0, 1].

10. The power generation prediction method based on cloud computing integration technology according to claim 1 is characterized in that: In S400, the method for predicting the power generation efficiency prediction value includes the following steps: S401. Collect microclimate simulation results and equipment aging prediction results; S402, constructing a prediction model: The input layer receives the multimodal time series data of microclimate simulation results and equipment aging prediction results as input; The convolutional layer is used to extract spatial features from the input data; The pooling layer is used to reduce data dimensions, reduce computational effort, and improve model efficiency; The LSTM layer is used to process the time series data after the convolution layer and capture long-term dependencies; The fully connected layer is used to integrate the information output by the LSTM layer and transform it into the final prediction output.

11. The power generation prediction method based on cloud computing integration technology according to claim 10 is characterized in that: S400 further includes the following steps: S403, compile model: Select Adam optimizer to compile the model; S404, power generation efficiency model training: Train the model using the training set data; S405, Power Generation Efficiency Model Evaluation: Use the test set to evaluate the model performance and calculate the error between the predicted power generation efficiency and the actual power generation efficiency; S406. Predicting power generation efficiency: Use the trained model to predict the new time series data and obtain the predicted value of power generation efficiency.

12. The power generation prediction method based on cloud computing integration technology according to claim 11, characterized in that: Convolution layer calculation, for each convolution kernel, the result of the convolution operation is calculated as:

13. The power generation prediction method based on cloud computing integration technology according to claim 12 is characterized in that: Pooling layer calculation, for the maximum pooling operation, the output is:

14. The power generation prediction method based on cloud computing integration technology according to claim 13 is characterized in that: Internal workings of LSTM cells; Forget gate calculation: f t =σ(W f ·[h t-1 , x t ]+b f ); Input gate calculation: i t =σ(W i ·[h t-1 , x t ]+b i ); Candidate cell state calculation: Current cell state calculation: Output gate calculation: O t =σ(W o ·[h t-1 , x t ]+b o ); Hidden state calculation: h t =o t *tanh(c t ).

15. The power generation prediction method based on cloud computing integration technology according to claim 14 is characterized in that: The predicted power generation efficiency is expressed as P=M(X), where M is the trained model, X is a set of feature vectors, and P is the predicted power generation efficiency.

Images