Prediction method for power generation flow data of hydraulic power plant

Abstract

The invention provides a hydraulic power plant power generation flow data prediction method, and relates to the technical field of hydraulic power plant power generation flow prediction, and the method comprises the steps: collecting meteorological data, landform data, soil characteristic data and hydraulic power plant operation data of a watershed where a hydraulic power plant is located through a multi-source sensor; data missing values are filled by adopting a long and short term memory network interpolation model, abnormal values in the data are detected by adopting a density-based spatial clustering method in combination with an isolated forest method, and the abnormal values are corrected by using a conversion-based smoothing algorithm; a deep belief network DBN and a convolutional neural network CNN are combined to construct a prediction model, data are input into the prediction model, feature extraction is performed on topographic and geomorphic data and soil characteristic data through the CNN, deep feature learning is performed on the extracted features, meteorological data and hydraulic power plant operation data through the DBN, and the power generation flow of the hydraulic power plant is predicted. By adopting the scheme, accurate and efficient prediction of the power generation flow of the hydraulic power plant is realized.

Description

Technical Field

[0001] This application relates to the field of hydropower plant power generation flow prediction technology, and in particular to a method and apparatus for predicting hydropower plant power generation flow data. Background Technology

[0002] Hydropower, as a clean and renewable energy source, plays a vital role in the energy structure. Accurate prediction of hydropower plant flow data is crucial for its efficient operation, water resource allocation, and stable power supply. Currently, predicting hydropower plant flow data faces numerous challenges. Traditional prediction methods rely on large amounts of high-quality historical data and complex mathematical models; however, prediction accuracy drops significantly when data is missing or abnormal. Existing technologies struggle to comprehensively consider the complex environment and operational characteristics of hydropower plants, resulting in predictions that fail to meet practical needs. Furthermore, external factors and human activities increase the uncertainty of hydropower plant flow, further complicating prediction. Therefore, developing accurate, efficient, and adaptable prediction methods is urgently needed. Summary of the Invention

[0003] This application aims to at least partially address one of the technical problems in the related art.

[0004] Therefore, the first objective of this application is to propose a method for predicting hydropower plant power generation flow data, which solves the problems of low prediction accuracy, great influence from data quality, and difficulty in comprehensively considering the influence of multiple factors in existing methods, and realizes accurate and efficient prediction of hydropower plant power generation flow data.

[0005] The second objective of this application is to propose a computer device.

[0006] The third objective of this application is to provide a non-transitory computer-readable storage medium.

[0007] To achieve the above objectives, the first aspect of this application proposes a method for predicting hydropower plant power generation flow data, comprising: collecting meteorological data, topographic data, soil characteristic data, and hydropower plant operation data of the watershed where the hydropower plant is located through multi-source sensors; filling missing values ​​in the data using a long short-term memory network interpolation model; detecting outliers in the data using a density-based spatial clustering method combined with an isolated forest method; and correcting outliers using a transformation-based smoothing algorithm; constructing a prediction model by combining a deep belief network (DBN) and a convolutional neural network (CNN); inputting the data into the prediction model; extracting features from the topographic data and soil characteristic data using the CNN; and performing deep feature learning on the extracted features, meteorological data, and hydropower plant operation data using the DBN to predict the hydropower plant power generation flow.

[0008] To achieve the above objectives, a second aspect of the present invention provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned method for predicting hydropower plant power generation flow data.

[0009] To achieve the above objectives, a third aspect of the present invention provides a non-transitory computer-readable storage medium that, when instructions in the storage medium are executed by a processor, enables the execution of a method for predicting hydropower plant power generation flow data.

[0010] This application employs advanced data processing methods, constructing a Long Short-Term Memory (LSTM) network interpolation model to predict and impute missing data. Compared to traditional simple interpolation methods, the model can fully learn the time-series characteristics of the data, more accurately reconstructing missing values ​​and avoiding information loss and prediction bias caused by missing data. Regarding outlier handling, it combines density-based spatial clustering and isolated forest algorithms. This combination can more accurately identify outliers, and the transformation-based smoothing algorithm can effectively correct outliers, ensuring data quality and stability and providing a reliable data foundation for subsequent predictions. Traditional methods are often not accurate enough in outlier handling and are easily affected by noise. This application introduces an early warning mechanism for abnormal events and automatically adjusts the power generation flow prediction model based on the type of abnormal event. Utilizing meteorological satellite data, hydrological monitoring data, and geographic information data, combined with a support vector machine algorithm, an abnormal event prediction model is constructed, enabling early prediction of natural disasters such as floods and droughts. When an abnormal event occurs, the parameters and structure of the prediction model are adjusted in a timely manner to better reflect the changing patterns of power generation flow under abnormal conditions. This ensures high prediction accuracy during abnormal events, providing strong support for the operation and scheduling of hydropower plants under extreme conditions and guaranteeing the safe and stable operation of hydropower plants. This application establishes a dynamic model update mechanism that monitors newly collected data in real time and automatically triggers incremental training when the data volume reaches a set threshold. This dynamic update method enables the model to adapt to changes in the watershed environment and the operating status of the hydropower plant in a timely manner, continuously learn new data features, and maintain good predictive performance. The model in this embodiment can continuously self-optimize, reduce manual intervention, improve the timeliness and accuracy of predictions, and better meet the needs of hydropower plants for real-time prediction of power generation flow in actual operation.

[0011] The method and apparatus for predicting hydropower plant power generation flow data proposed in this application organically combine data quality processing methods, multi-source data fusion technology, deep learning algorithms, abnormal event early warning and model adjustment mechanisms, as well as dynamic update and error correction mechanisms. This significantly improves the prediction accuracy and adaptability of hydropower plant power generation flow data, providing a reliable guarantee for the scientific scheduling and efficient operation of hydropower plants.

[0012] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0013] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating a method for predicting hydropower generation flow data provided in Embodiment 1 of this application. Detailed Implementation

[0014] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0015] The following describes a method and apparatus for predicting hydropower plant flow data according to embodiments of this application, with reference to the accompanying drawings.

[0016] Figure 1 This is a flowchart illustrating a method for predicting hydropower generation flow data provided in Embodiment 1 of this application.

[0017] like Figure 1 As shown, the method for predicting the power generation flow data of this hydropower plant includes the following steps: Step 101: Collect meteorological data, topographic data, soil characteristic data, and hydropower plant operation data of the watershed where the hydropower plant is located through multi-source sensors; In this embodiment, after data collection, the data is preprocessed, including: A denoising algorithm based on wavelet transform is used to remove noise, and the data is normalized using the minimum-maximum normalization method. Key feature variables are then selected by calculating the information gain ratio between the feature variables and the power generation flow.

[0018] Step 102: Use a long short-term memory network interpolation model to fill in missing values ​​in the data, use a density-based spatial clustering method combined with an isolated forest method to detect outliers in the data, and use a transformation-based smoothing algorithm to correct outliers. In this embodiment, the forgetting gate of the Long Short-Term Memory network interpolation model The calculation formula is:

[0019] in, It is the sigmoid activation function. It is the forget gate weight matrix. This indicates that the previous state will be hidden. Input at the current time To splice, It is the forget gate bias.

[0020] In this embodiment, a density-based spatial clustering method combined with an isolated forest method is used to detect outliers in the data, including: Density-based spatial clustering is used to perform cluster analysis on the data to identify outliers. The Isolation Forest algorithm is then used to locate outliers within these outliers. In density-based spatial clustering methods, for points in the dataset its neighborhood The calculation formula is:

[0021] in, For the dataset, Point and points The distance between them; like ,Sure As the core point; In the Isolation Forest algorithm, data points are calculated. The isolated forest score is represented as:

[0022] in, It is a point Path length, It is related to the size of the dataset. Regarding the average path length, hour, , yes Harmonic number.

[0023] Step 103: A prediction model is constructed by combining Deep Belief Network (DBN) and Convolutional Neural Network (CNN). The data is input into the prediction model. The CNN extracts features from the topographic data and soil property data. The DBN performs deep feature learning on the extracted features, meteorological data, and hydropower plant operation data to predict the hydropower plant's power generation flow.

[0024] In this embodiment, a prediction model is constructed by combining a Deep Belief Network (DBN) and a Convolutional Neural Network (CNN), including: The CNN-DBN model is built using the TensorFlow framework. The CNN includes convolutional layers and pooling layers, and the DBN includes a restricted Boltzmann machine. Training the prediction model includes: The model is trained using a large amount of historical data, the model parameters are optimized using a stochastic gradient descent algorithm with an adaptive learning rate, regularization techniques are introduced to prevent overfitting, and the model is evaluated using cross-validation. The regularization techniques include L1 and L2 regularization. Training the prediction model further includes: The system monitors the number of collected data samples in real time and automatically triggers a model update process when the number of data samples reaches a threshold, using the new data to incrementally train the model. For example, the data update threshold is set to add 100 data samples, and incremental model training is triggered when this threshold is reached.

[0025] In this embodiment, the prediction results are also verified and corrected, specifically: The prediction results are evaluated using root mean square error, mean absolute error, and coefficient of determination. If the preset accuracy requirements are not met, the prediction results are corrected by analyzing the distribution characteristics and trends of the prediction error, combined with the actual operation of the hydropower plant and expert experience.

[0026] In this embodiment, abnormal event prediction and model adjustment are also included, specifically: An early warning mechanism for abnormal events is introduced. Meteorological satellite data, hydrological monitoring data, and geographic information data are used, combined with a support vector machine algorithm to construct an abnormal event prediction model. When an abnormal event is predicted, the parameters and structure of the power generation flow prediction model are automatically adjusted. The construction of the abnormal event prediction model includes: For linearly separable problems, solve the optimization problem:

[0027] The constraints are set as follows:

[0028] in, It is the input sample. These are sample labels. It is the sample size; For nonlinearly separable problems, a kernel function is introduced. such as radial basis kernel function , These are kernel function parameters.

[0029] This application employs advanced data processing methods, constructing a Long Short-Term Memory (LSTM) network interpolation model to predict and impute missing data. Compared to traditional simple interpolation methods, the model can fully learn the time-series characteristics of the data, more accurately reconstructing missing values ​​and avoiding information loss and prediction bias caused by missing data. Regarding outlier handling, it combines density-based spatial clustering and isolated forest algorithms. This combination can more accurately identify outliers, and the transformation-based smoothing algorithm can effectively correct outliers, ensuring data quality and stability and providing a reliable data foundation for subsequent predictions. Traditional methods are often not accurate enough in outlier handling and are easily affected by noise. This application introduces an early warning mechanism for abnormal events and automatically adjusts the power generation flow prediction model based on the type of abnormal event. Utilizing meteorological satellite data, hydrological monitoring data, and geographic information data, combined with a support vector machine algorithm, an abnormal event prediction model is constructed, enabling early prediction of natural disasters such as floods and droughts. When an abnormal event occurs, the parameters and structure of the prediction model are adjusted in a timely manner to better reflect the changing patterns of power generation flow under abnormal conditions. This ensures high prediction accuracy during abnormal events, providing strong support for the operation and scheduling of hydropower plants under extreme conditions and guaranteeing the safe and stable operation of hydropower plants. This application establishes a dynamic model update mechanism that monitors newly collected data in real time and automatically triggers incremental training when the data volume reaches a set threshold. This dynamic update method enables the model to adapt to changes in the watershed environment and the operating status of the hydropower plant in a timely manner, continuously learn new data features, and maintain good predictive performance. The model in this embodiment can continuously self-optimize, reduce manual intervention, improve the timeliness and accuracy of predictions, and better meet the needs of hydropower plants for real-time prediction of power generation flow in actual operation.

[0030] The method and apparatus for predicting hydropower plant power generation flow data proposed in this application organically combine data quality processing methods, multi-source data fusion technology, deep learning algorithms, abnormal event early warning and model adjustment mechanisms, as well as dynamic update and error correction mechanisms. This significantly improves the prediction accuracy and adaptability of hydropower plant power generation flow data, providing a reliable guarantee for the scientific scheduling and efficient operation of hydropower plants.

[0031] This embodiment takes Hydropower Plant A, located in a mountainous area, as the subject of the example. The watershed where this hydropower plant is located has complex terrain, variable meteorological conditions, and data gaps and anomalies are relatively common. Therefore, the prediction is carried out according to the following steps: (1) Data Acquisition and Preprocessing: Multiple meteorological monitoring stations, topographic survey points, and soil sampling points were deployed in the watershed where Hydropower Plant A is located. The meteorological monitoring stations collected real-time data on rainfall, temperature, wind speed, humidity, etc.; topographic data were obtained using UAVs equipped with high-precision topographic mapping equipment; soil samples were collected in different areas, and soil texture, porosity, and other characteristics were analyzed in the laboratory. At the same time, real-time operational data such as the power generation capacity and water level changes of the hydropower plant were collected.

[0032] The collected data is transmitted wirelessly to the data processing center. First, a wavelet transform-based denoising algorithm is used to reduce noise in the data. Taking rainfall data as an example, wavelet transform is used to decompose it into different frequency sub-bands. It is found that there are a large number of noise coefficients in the high-frequency sub-bands. After setting an appropriate threshold to remove these noise coefficients, inverse wavelet transform is then used to reconstruct the rainfall data, effectively removing noise interference. Next, the min-max normalization method is used to normalize all data to the [0,1] interval. For example, the maximum power generation during a certain period is... The minimum value is Normalized power generation Finally, the information gain ratio of each characteristic variable to the power generation flow was calculated, and key characteristic variables such as rainfall, water level change, and soil porosity were screened out.

[0033] (2) Handling Missing Data and Outliers: When analyzing the power generation flow data of Hydropower Plant A, it was found that there were missing data in some time periods. An LSTM interpolation model was constructed to handle the missing data. When training the LSTM model, parameters such as the forget gate, input gate, and output gate were set. After multiple rounds of training, the model learned the time series characteristics of the power generation flow data. For example, for a certain period of missing 3 days of power generation flow data, the model accurately predicted and filled in the missing values ​​based on the changing trend of the data before and after.

[0034] Forget gate in Long Short-Term Memory Network Interpolation Model The calculation formula is:

[0035] in, It is the sigmoid activation function. It is the forget gate weight matrix. This indicates that the previous state will be hidden. Input at the current time To splice, It is the forget gate bias.

[0036] For outlier handling, the density-based spatial clustering algorithm DBSCAN is used to perform cluster analysis on the data. A neighborhood radius is set. Minimum number of samples Outliers were initially identified. The Isolation Forest algorithm was then used to further pinpoint outliers. For detected outliers, a combination of density-based spatial clustering and the Isolation Forest algorithm was employed for detection, followed by a transformation-based smoothing algorithm for correction.

[0037] In density-based spatial clustering algorithms, for points in the dataset its neighborhood The calculation formula is:

[0038] like ,but As the core point, among which It is a dataset. Point and points The distance between points is MinPts, which is the minimum number of points.

[0039] In the Isolation Forest algorithm, data points are calculated. Isolated Forest Score:

[0040] in It is a point Path length, It is related to the size of the dataset. Regarding the average path length, when hour, , yes Harmonic number.

[0041] (3) Constructing a multi-factor fusion prediction model: A CNN-DBN model was built using the TensorFlow framework. In the CNN part, three convolutional layers and two pooling layers were designed, with convolutional kernel sizes of 3×3 and 5×5, and a stride of 1. The pooling layers used max pooling with a pooling window size of 2×2. Feature extraction was performed on topographic and soil property data through these layers. The extracted features were input into the DBN along with meteorological data and hydropower plant operation data. The DBN contained three layers of restricted Boltzmann machines (RBMs), with 128, 64, and 32 neurons in each RBM layer, respectively. After multiple rounds of training, the model gradually converged and was able to effectively learn the complex relationship between various factors and power generation flow.

[0042] (4) Abnormal event prediction and model adjustment: Collect meteorological satellite data, hydrological monitoring data and geographic information data of the basin where Hydropower Plant A is located, and construct an abnormal event prediction model based on the support vector machine prediction algorithm.

[0043] When using support vector machines to predict anomalous events, for the linearly separable case, an optimization problem is solved. The constraints are:

[0044] in It is the input sample. These are sample labels. It is the number of samples; for nonlinearly separable problems, a kernel function is introduced. such as radial basis kernel function , These are kernel function parameters.

[0045] Because the data in this region exhibits non-linear separability, a radial basis kernel function is chosen. ,set up After training the support vector machine prediction model, it can accurately predict abnormal events such as floods and droughts. When a flood is predicted, the power generation flow prediction model is adjusted by increasing the weight of relevant features such as rainfall and river water level by 2096 and expanding the prediction interval. When a drought is predicted, the weight of features related to rainfall is reduced by 30%, features related to water resource reserves are increased, and the model is retrained.

[0046] (5) Model Training and Optimization: The model was trained using historical data from Hydropower Plant A over the past 5 years. An adaptive learning rate stochastic gradient descent algorithm was employed, with an initial learning rate of... attenuation coefficient During training, the learning rate is dynamically adjusted based on the model's training error. Simultaneously, L1 and L2 regularization techniques are introduced, with an L1 regularization coefficient of 0.001 and an L2 regularization coefficient of 0.01, effectively preventing overfitting. Five-fold cross-validation is used to evaluate and fine-tune the model, and the model parameters that perform best on the validation set are selected as the final model parameters.

[0047] (6) Dynamically Update the Prediction Model: Real-time monitoring of newly collected data from Hydropower Plant A, with a data update threshold set at 100 new data samples. When this threshold is reached, the model update process is automatically triggered. An online learning algorithm is used to incrementally train the model with new data, enabling the model to adapt to changes in the watershed environment and the hydropower plant's operating status in a timely manner. For example, when a small reservoir is newly built in the watershed, affecting water flow convergence, the new data reflects this change, and the model adjusts its parameters through incremental training to maintain the accuracy of the prediction.

[0048] (7) Validation and Correction of Prediction Results: The prediction results are evaluated using validation indicators such as root mean square error, mean absolute error, and coefficient of determination. If the root mean square error exceeds the preset threshold of 0.5, the distribution of prediction errors is analyzed. If the prediction error is found to be large under heavy rain conditions, a linear regression error correction model based on rainfall is established. ,in For rainfall, the regression coefficients were determined using the least squares method. and This allows for the correction of prediction results, thereby improving the accuracy of predictions.

[0049] To implement the above embodiments, the present invention also proposes a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the methods described in the above embodiments.

[0050] To implement the above embodiments, the present invention also proposes a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method of the above embodiments.

[0051] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0052] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0053] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0054] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0055] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0056] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0057] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0058] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for predicting hydropower plant power generation flow data, characterized in that, include: Meteorological data, topographic data, soil characteristic data, and hydropower plant operation data of the watershed where the hydropower plant is located are collected through multi-source sensors. A long short-term memory network interpolation model is used to fill in missing values ​​in the data, a density-based spatial clustering method combined with an isolated forest method is used to detect outliers in the data, and a transformation-based smoothing algorithm is used to correct outliers. A prediction model is constructed by combining Deep Belief Network (DBN) and Convolutional Neural Network (CNN). Data is input into the prediction model, and CNN is used to extract features from topographic and soil property data. DBN is then used to perform deep feature learning on the extracted features, meteorological data, and hydropower plant operation data to predict the power generation flow of the hydropower plant.

2. The method as described in claim 1, characterized in that, Before imputing missing values ​​in the data, the following steps are also included: A denoising algorithm based on wavelet transform is used to remove data noise; The data is normalized using the min-max normalization method; Calculate the information gain ratio between the characteristic variables and the power generation flow, determine the key characteristic variables based on the information gain ratio, and screen out the data related to the key characteristic variables.

3. The method as described in claim 1, characterized in that, The forgetting gate in the Long Short-Term Memory Network interpolation model The calculation formula is: in, It is the sigmoid activation function. It is the forget gate weight matrix. This indicates that the previous state will be hidden. Input at the current time To splice, It is the forget gate bias.

4. The method as described in claim 1, characterized in that, The method of using density-based spatial clustering combined with the isolated forest method to detect outliers in the data includes: Density-based spatial clustering is used to perform cluster analysis on the data to identify outliers. The Isolation Forest algorithm is then used to locate outliers within these outliers. In density-based spatial clustering methods, for points in the dataset its neighborhood The calculation formula is: in, For the dataset, Point and points The distance between them; like ,Sure As the core point; In the Isolation Forest algorithm, data points are calculated. The isolated forest score is represented as: in, It is a point Path length, It is related to the size of the dataset. Regarding the average path length, hour, , yes Harmonic number.

5. The method as described in claim 1, characterized in that, A prediction model is constructed by combining a Deep Belief Network (DBN) with a Convolutional Neural Network (CNN), including: The CNN-DBN model is built using the TensorFlow framework. The CNN includes convolutional layers and pooling layers, and the DBN includes a restricted Boltzmann machine. Training the prediction model includes: The model is trained using historical data, the model parameters are optimized using a stochastic gradient descent algorithm with an adaptive learning rate, regularization techniques are introduced to prevent overfitting, and the model is evaluated using cross-validation. The regularization techniques include L1 and L2 regularization. Training the prediction model further includes: The number of data samples collected is monitored in real time, and when the number of data samples reaches a threshold, the model is incrementally trained using new data.

6. The method as described in claim 1, characterized in that, The method further includes: The prediction results are evaluated using root mean square error, mean absolute error, and coefficient of determination. If the preset accuracy requirements are not met, the prediction results are corrected by analyzing the distribution characteristics and trends of the prediction error, combined with the actual operation of the hydropower plant and expert experience.

7. The method as described in claim 1, characterized in that, The method further includes: An early warning mechanism for abnormal events is introduced. Meteorological satellite data, hydrological monitoring data, and geographic information data are used to construct an abnormal event prediction model by combining support vector machine algorithm. When an abnormal event is predicted to occur, the structure and parameters of the prediction model are adjusted.

8. The method as described in claim 7, characterized in that, Constructing an anomaly event prediction model includes: For linearly separable problems, solve the optimization problem: The constraints are set as follows: in, It is the input sample. These are sample labels. It is the sample size; For nonlinearly separable problems, a kernel function is introduced. .

9. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method as described in any one of claims 1-8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-8.

Images