A reservoir flood control risk prediction method and system

Abstract

The present invention relates to the field of water conservancy engineering and disaster prevention and mitigation technology, and in particular to a reservoir flood risk prediction method and system. The steps are as follows: collect hydrological, meteorological and geological data in real time through sensors, and form a data set based on the collected data; transmit the data in the data set, and strengthen the data security through blockchain encryption technology during the transmission process; transmit the pre-processed data to the data processing center, and the data processing center pre-processes and analyzes the received data to generate flood risk prediction results and early warning information. This party realizes real-time prediction of reservoir flood risk through multi-source data fusion and intelligent algorithms, and provides scientific flood control decision support.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy engineering and disaster prevention and mitigation, and in particular to a reservoir flood control risk prediction method and system. Background Art

[0002] Reservoirs play an important role in flood prevention and disaster reduction, but due to climate change and the increase in extreme weather events, the flood control pressure faced by reservoirs is also gradually increasing. How to use a variety of environmental monitoring data to predict flood control risk rates is a major problem that needs to be solved urgently, which directly affects the decision-making of water conservancy and other departments in dealing with flood disasters. Traditional flood risk assessment methods usually rely on historical data and experience, are difficult to adapt to complex and changeable climate environments, and lack real-time performance and accuracy. By integrating a variety of sensor data and using transformer-based LSTM technology, the present invention can achieve more accurate and real-time flood risk prediction, improve the flood control capacity of reservoirs, and help water conservancy and other departments improve their decision-making level in dealing with flood disasters and reduce the impact of disasters.

[0003] Therefore, the present invention provides a reservoir flood control risk prediction method and system to improve the reservoir flood control management level and ensure flood control safety. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention develops a reservoir flood risk prediction method and system, the main purpose of which is to achieve real-time prediction of reservoir flood risk through multi-source data fusion and intelligent algorithm, and provide scientific flood control decision support.

[0005] The technical solution to the technical problem of the present invention is a reservoir flood control risk prediction method, comprising the following steps:

[0006] S1. Data collection: collect hydrological, meteorological and geological data in real time through sensors, and form a data set based on the collected data;

[0007] S2. Data transmission: The data in the dataset is transmitted, and the data security is enhanced through blockchain encryption technology during the transmission process;

[0008] S3. Data analysis: The preprocessed data is transmitted to the data processing center, which preprocesses and analyzes the received data to generate flood risk prediction results and warning information.

[0009] In a specific implementation manner, S1 is as follows:

[0010] Through sensors in reservoir management centers, meteorological monitoring stations, hydrological monitoring stations and geological monitoring stations Interval duration in time Collect hydrological, meteorological and geological data, including reservoir water level , outflow , evapotranspiration , rainfall , meteorological data , earthquake activity data and geological deformation data , meteorological data These include air temperature, humidity and wind speed;

[0011] Build a dataset based on the collected data , , , , , , dataset Communist Party of China group data, data set The total time of data collection is The time interval for collecting each set of data is , Indicated in The first Group data, , , , , , and Respectively expressed in The first The group includes reservoir water levels, outflow, evapotranspiration, rainfall, meteorological data, seismic activity data and geological deformation data.

[0012] In a specific implementation manner, S2 is as follows:

[0013] S2.1. Blockchain network deployment:

[0014] Treat the reservoir management center, meteorological monitoring station, hydrological monitoring station, and geological monitoring station as separate nodes in the blockchain, and choose private chain as the blockchain network architecture to ensure data security and privacy;

[0015] S2.2. Data processing before uploading to the chain:

[0016] The dataset Each type of data is packaged into data blocks at a certain time interval. Suppose the data collection time is , the time Divide evenly into time intervals, there are data blocks, data block collection Expressed as , Indicates Data blocks, hash each data block, generate a unique hash value to ensure data integrity;

[0017] Reuse The symmetric encryption algorithm encrypts the data block, and the encryption key is , the specific calculation is as follows:

[0018] ,

[0019] in, Indicates that the encryption key is Symmetric encryption algorithm, Represents a collection of data blocks The data in the data block is symmetrically encrypted;

[0020] S2.3. Data upload and consensus mechanism:

[0021] The symmetrically encrypted data The hash value of each data block is uploaded to the nodes in the blockchain network in turn, and each node verifies and confirms the uploaded data through the consensus mechanism of the blockchain.

[0022] In a specific implementation manner, S3 is as follows:

[0023] The data processing center receives the symmetrically encrypted data After calculating the hash value of each data block, the received data is first restored to a data set The initial data collected in the process is then preprocessed for the restored data. The data in the data is preprocessed, and then the preprocessed data is analyzed through big data and based on the self-attention mechanism model Long short-term memory network for R The algorithm performs data analysis to obtain comprehensive flood risk prediction results, and finally conducts risk assessment to generate early warning information.

[0024] S3.1. Data preprocessing:

[0025] Data cleaning: remove data sets Outliers and noise data in the data;

[0026] Data smoothing: Use the sliding average method to smooth the data set. The data is smoothed, and the specific calculation is as follows:

[0027] ,

[0028] in, Represents the original data smooth The data after the item, represents the sliding window size, After smoothing Group data, , , , , , , and They represent the smoothed reservoir water level, outflow, evapotranspiration, rainfall, meteorological data, seismic activity data and geological deformation data respectively.

[0029] S3.2. Predict hydrological, meteorological, geological and comprehensive flood control risks through data analysis:

[0030] S3.2.1. Hydrological prediction:

[0031] To predict the water level change of the reservoir, the specific calculation is as follows:

[0032] ,

[0033] in, The predicted The water level of the reservoir, express The actual measured time The water level of the reservoir, , and Respectively represent the actual measured Items of rainfall, dispersion and outflow, represents the reservoir area;

[0034] S3.2.2 Weather forecast:

[0035] The time series analysis method is used to predict the meteorological data. Specifically, the meteorological data is predicted by the autoregressive moving average model ARIMA. The calculation formula is as follows:

[0036] ,

[0037] in, The predicted Items of meteorological data, , represents the constant term of the ARIMA formula, Respectively represent the actual measured , ,…, Items of meteorological data, Indicates The autoregression coefficients are The autoregressive coefficients, , Indicates The moving average coefficients are The moving average coefficient, Indicates Item white noise;

[0038] S3.2.3, Geological prediction:

[0039] Based on Network and The model's geological hazard risk assessment model evaluates seismic activity and geological deformation. The calculation process is as follows:

[0040] ,

[0041] in, Indicates Item geological risk score, Indicates based on Network and Model of geological disaster risk assessment, Indicates geological deformation data, Indicates Seismic activity data.

[0042] S3.2.4. Comprehensive flood risk forecast:

[0043] After smoothing each item Respectively , and Combined into new data , using a Long-term and short-term networks The algorithm predicts the risk rate;

[0044] use Building a subsequence temporal representation learner ,pass Learning compressed and contextual subsequence representations from long time series, Use mask reconstruction task, and then pass random mask data 75% of the data is masked ,pass right Perform reconstruction optimization to obtain the mask subsequence , and by reconstructing the loss function right For optimization, the calculation process is as follows:

[0045] ,

[0046] From the data The remaining 25% of the data is not masked Reconstruct the unmasked subsequence in the temporal context of , and by reconstructing the loss function right For optimization, the calculation process is as follows:

[0047] ,

[0048] Masked subsequence and unmasked subsequence Together they constitute the time representation of the processed subsequence ;

[0049] (1) Long-term trend extraction:

[0050] From the subsequence time representation To capture long-term trend features, we use stacked one-dimensional dilated convolutional layers to expand the temporal receptive field and capture long-term trends. Layer, the one-dimensional dilated convolution operation is as follows:

[0051] ,

[0052] in, represents the output of the first layer of the one-dimensional dilated convolutional layer, Represents subsequence time representation The elements, C represents the convolution kernel of a one-dimensional dilated convolution operation, represents the convolution kernel size, represents the variable in the accumulation operation, represents the expansion rate, Represents a one-dimensional dilated convolution operation;

[0053] After obtaining the output of the first layer of the one-dimensional dilated convolutional layer, the result of the previous layer is convolved with the convolution kernel of the next layer, and then the output of the next layer is obtained through activation function and maximum pooling operation. The specific operations are as follows:

[0054] ,

[0055] …

[0056] ,

[0057] …

[0058] ,

[0059] in, Indicates The output of the layer, , express The operation of the activation function, Represents the maximum pooling, the output of the last layer Recorded as long-term trend characteristics ;

[0060] (2) Short-term trend extraction:

[0061] Capturing local fine-grained trend features in short-term historical sequences, using existing spatiotemporal graph neural networks As a short-term trend extractor, to extract short-term trend features, the calculation process is as follows:

[0062] ,

[0063] in, Indicates short-term trend characteristics. represents the adjacency matrix;

[0064] (3) Feature fusion and risk prediction:

[0065] The long-term trend features and short-term trend features are integrated to obtain the final prediction results. The feature integration and prediction calculation process is as follows:

[0066] ,

[0067] in, express of Iteration operation, Represents the prediction result, represents a multi-layer perceptron, Represents a join operation.

[0068] The present invention also provides a reservoir flood control risk prediction system, which implements a reservoir flood control risk prediction method, specifically including the following modules:

[0069] (1) Data acquisition module:

[0070] Hydrological monitoring equipment: water level meters and flow meters are deployed at different locations of the reservoir to monitor water level and flow changes in real time;

[0071] Meteorological monitoring equipment: Rain gauges, temperature and humidity sensors, and anemometers are installed around the reservoir and in the upstream area to collect rainfall, temperature, humidity, and wind speed;

[0072] Geological monitoring equipment: seismometers, inclinometers, and strain gauges are deployed on the reservoir dam and surrounding areas to monitor seismic activity and geological deformation;

[0073] (2) Data transmission module: transmits the pre-processed data, and uses blockchain encryption technology to enhance data security during the transmission process;

[0074] (3) Data analysis module: The preprocessed data is transmitted to the data processing center, which stores and analyzes the received data to generate flood risk prediction results and warning information.

[0075] The effects provided in the content of the invention are only the effects of the embodiments, rather than all the effects of the invention. The above technical solution has the following advantages or beneficial effects:

[0076] The present invention provides an efficient and accurate reservoir flood risk prediction method and its implementation system, which realizes real-time prediction of flood risk through automation and intelligent technology, and provides a scientific flood control management solution. The system has the advantages of real-time monitoring, high precision, automation and intelligent early warning, and can be widely used in the field of reservoir flood control management, providing strong support for improving reservoir flood control capacity and ensuring flood control safety. It innovatively introduces transformer-based LSTM algorithm, Internet of Things technology and blockchain technology, further improving the accuracy, security and transparency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0078] Figure 1 It is a flow chart of the method of the present invention.

[0079] Figure 2 The following is a line chart comparing the accuracy of different methods. DETAILED DESCRIPTION

[0080] In order to clearly illustrate the technical features of the present invention, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0081] Example 1

[0082] Depend on Figure 1 As shown in the figure, a reservoir flood control risk prediction method is as follows:

[0083] S1. Data collection: collect hydrological, meteorological and geological data in real time through sensors, and form a data set based on the collected data;

[0084] S2. Data transmission: The data in the dataset is transmitted, and the data security is enhanced through blockchain encryption technology during the transmission process;

[0085] S3. Data analysis: The preprocessed data is transmitted to the data processing center, which preprocesses and analyzes the received data to generate flood risk prediction results and warning information.

[0086] In a specific implementation manner, S1 is as follows:

[0087] Through sensors in reservoir management centers, meteorological monitoring stations, hydrological monitoring stations and geological monitoring stations Interval duration in time Collect hydrological, meteorological and geological data, including reservoir water level , outflow , evapotranspiration , rainfall , meteorological data , earthquake activity data and geological deformation data , meteorological data These include air temperature, humidity and wind speed;

[0088] Build a dataset based on the collected data , , , , , , dataset Communist Party of China group data, data set The total time of data collection is The time interval for collecting each set of data is , Indicated in The first Group data, , , , , , and Respectively expressed in The first The group includes reservoir water levels, outflow, evapotranspiration, rainfall, meteorological data, seismic activity data and geological deformation data.

[0089] In a specific implementation manner, S2 is as follows:

[0090] S2.1. Blockchain network deployment:

[0091] Treat the reservoir management center, meteorological monitoring station, hydrological monitoring station, and geological monitoring station as separate nodes in the blockchain, and choose private chain as the blockchain network architecture to ensure data security and privacy;

[0092] S2.2. Data processing before uploading to the chain:

[0093] The dataset Each type of data is packaged into data blocks at a certain time interval. Suppose the data collection time is , the time Divide evenly into time intervals, there are data blocks, data block collection Expressed as , Indicates Data blocks, hash each data block, generate a unique hash value to ensure data integrity;

[0094] Reuse The symmetric encryption algorithm encrypts the data block, and the encryption key is , the specific calculation is as follows:

[0095] ,

[0096] in, Indicates that the encryption key is Symmetric encryption algorithm, Represents a collection of data blocks The data in the data block is symmetrically encrypted;

[0097] S2.3. Data upload and consensus mechanism:

[0098] The symmetrically encrypted data The hash value of each data block is uploaded to the nodes in the blockchain network in turn, and each node verifies and confirms the uploaded data through the consensus mechanism of the blockchain.

[0099] In a specific implementation manner, S3 is as follows:

[0100] The data processing center receives the symmetrically encrypted data After calculating the hash value of each data block, the received data is first restored to a data set The initial data collected in the process is then preprocessed for the restored data. The data in the data is preprocessed, and then the preprocessed data is analyzed through big data and based on the self-attention mechanism model Long short-term memory network for R The algorithm performs data analysis to obtain comprehensive flood risk prediction results, and finally conducts risk assessment to generate early warning information.

[0101] S3.1. Data preprocessing:

[0102] Data cleaning: remove data sets Outliers and noise data in the data;

[0103] Data smoothing: Use the sliding average method to smooth the data set. The data is smoothed, and the specific calculation is as follows:

[0104] ,

[0105] in, Represents the original data smooth The data after the item, represents the sliding window size, After smoothing Group data, , , , , , , and They represent the smoothed reservoir water level, outflow, evapotranspiration, rainfall, meteorological data, seismic activity data and geological deformation data respectively.

[0106] S3.2. Predict hydrological, meteorological, geological and comprehensive flood control risks through data analysis:

[0107] S3.2.1. Hydrological prediction:

[0108] To predict the water level change of the reservoir, the specific calculation is as follows:

[0109] ,

[0110] in, The predicted The water level of the reservoir, express The actual measured time The water level of the reservoir, , and Respectively represent the actual measured Items of rainfall, dispersion and outflow, represents the reservoir area;

[0111] S3.2.2 Weather forecast:

[0112] The time series analysis method is used to predict the meteorological data. Specifically, the meteorological data is predicted by the autoregressive moving average model ARIMA. The calculation formula is as follows:

[0113] ,

[0114] in, The predicted Items of meteorological data, , represents the constant term of the ARIMA formula, Respectively represent the actual measured , ,…, Items of meteorological data, Indicates The autoregression coefficients are The autoregressive coefficients, , Indicates The moving average coefficients are The moving average coefficient, Indicates Item white noise;

[0115] S3.2.3, Geological prediction:

[0116] Based on Network and The model's geological hazard risk assessment model evaluates seismic activity and geological deformation. The calculation process is as follows:

[0117] ,

[0118] in, Indicates Item geological risk score, Indicates based on Network and Model of geological disaster risk assessment, Indicates geological deformation data, Indicates Seismic activity data.

[0119] S3.2.4. Comprehensive flood risk forecast:

[0120] After smoothing each item Respectively , and Combined into new data , using a Long-term and short-term networks The algorithm predicts the risk rate;

[0121] use Building a subsequence temporal representation learner ,pass Learning compressed and contextual subsequence representations from long time series, Use mask reconstruction task, and then pass random mask data 75% of the data is masked ,pass right Perform reconstruction optimization to obtain the mask subsequence , and by reconstructing the loss function right For optimization, the calculation process is as follows:

[0122] ,

[0123] From the data The remaining 25% of the data is not masked Reconstruct the unmasked subsequence in the temporal context of , and by reconstructing the loss function right For optimization, the calculation process is as follows:

[0124] ,

[0125] Masked subsequence and unmasked subsequence Together they constitute the time representation of the processed subsequence ;

[0126] (1) Long-term trend extraction:

[0127] From the subsequence time representation To capture long-term trend features, we use stacked one-dimensional dilated convolutional layers to expand the temporal receptive field and capture long-term trends. Layer, the one-dimensional dilated convolution operation is as follows:

[0128] ,

[0129] in, represents the output of the first layer of the one-dimensional dilated convolutional layer, Represents subsequence time representation The elements, C represents the convolution kernel of a one-dimensional dilated convolution operation, represents the convolution kernel size, represents the variable in the accumulation operation, represents the expansion rate, Represents a one-dimensional dilated convolution operation;

[0130] After obtaining the output of the first layer of the one-dimensional dilated convolutional layer, the result of the previous layer is convolved with the convolution kernel of the next layer, and then the output of the next layer is obtained through activation function and maximum pooling operation. The specific operations are as follows:

[0131] ,

[0132] …

[0133] ,

[0134] …

[0135] ,

[0136] in, Indicates The output of the layer, , express The operation of the activation function, Represents the maximum pooling, the output of the last layer Recorded as long-term trend characteristics ;

[0137] (2) Short-term trend extraction:

[0138] Capturing local fine-grained trend features in short-term historical sequences, using existing spatiotemporal graph neural networks As a short-term trend extractor, to extract short-term trend features, the calculation process is as follows:

[0139] ,

[0140] in, Indicates short-term trend characteristics. represents the adjacency matrix;

[0141] (3) Feature fusion and risk prediction:

[0142] The long-term trend features and short-term trend features are integrated to obtain the final prediction results. The feature integration and prediction calculation process is as follows:

[0143] ,

[0144] in, express of Iteration operation, Represents the prediction result, represents a multi-layer perceptron, Represents a join operation.

[0145] Example 2

[0146] A reservoir flood control risk prediction system implements a reservoir flood control risk prediction method, specifically including the following modules:

[0147] (1) Data acquisition module:

[0148] Hydrological monitoring equipment: water level meters and flow meters are deployed at different locations of the reservoir to monitor water level and flow changes in real time;

[0149] Meteorological monitoring equipment: Rain gauges, temperature and humidity sensors, and anemometers are installed around the reservoir and in the upstream area to collect rainfall, temperature, humidity, and wind speed;

[0150] Geological monitoring equipment: seismometers, inclinometers, and strain gauges are deployed on the reservoir dam and surrounding areas to monitor seismic activity and geological deformation;

[0151] (2) Data transmission module: transmits the pre-processed data, and uses blockchain encryption technology to enhance data security during the transmission process;

[0152] (3) Data analysis module: The preprocessed data is transmitted to the data processing center, which stores and analyzes the received data to generate flood risk prediction results and warning information.

[0153] Example 3

[0154] As shown in Table 1, in order to further verify the beneficial effects of the present invention, the present invention is compared with the existing method in terms of prediction accuracy based on the data set collected by the present invention. The calculation results of the root mean square error, mean absolute error and mean absolute percentage error show that the present invention has a small error and a higher prediction accuracy. Figure 2 It can be seen intuitively that the accuracy of the present invention is higher.

[0155] Table 1 Comparison results of this method with other methods

[0156]

[0157] Although the above describes the specific implementation mode of the invention in conjunction with the drawings, it is not intended to limit the scope of protection of the invention. Based on the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present invention.

Claims

1. A reservoir flood risk prediction method, characterized in that: The following steps are involved: S1. Data collection: collect hydrological, meteorological and geological data in real time through sensors, and form a data set based on the collected data ; S2. Data transfer: Data set The data in the network is transmitted, and the data security is enhanced through blockchain encryption technology during the transmission process; S3, data analysis: the pre-processed data is transmitted to the data processing center, which pre-processes and analyzes the received data to generate flood risk prediction results and early warning information; S3 is as follows: S3.

1. Data preprocessing: Data cleaning: remove data sets Outliers and noise data in the data; Data smoothing: Use the sliding average method to smooth the data set. The data is smoothed, and the specific calculation is as follows: ， in, Representation dataset The original data in smooth The data after the item, represents the sliding window size, After smoothing Group data, , , , , , , and They represent the smoothed reservoir water level, outflow, evapotranspiration, rainfall, meteorological data, seismic activity data and geological deformation data respectively; S3.

2. Predict hydrological, meteorological, geological and comprehensive flood control risks through data analysis: S3.2.

1. Hydrological prediction: To predict the water level change of the reservoir, the specific calculation is as follows: ， in, The predicted The water level of the reservoir, express The actual measured time The water level of the reservoir, , and Respectively represent the actual measured Items of rainfall, dispersion and outflow, represents the reservoir area; S3.2.2 Weather forecast: The time series analysis method is used to predict the meteorological data. Specifically, the meteorological data is predicted by the autoregressive moving average model ARIMA. The calculation formula is as follows: ， in, The predicted Items of meteorological data, , represents the constant term of the ARIMA formula, Respectively represent the actual measured , ,…, Items of meteorological data, Indicates The autoregression coefficients are The autoregressive coefficients, , Indicates The moving average coefficients are The moving average coefficient, Indicates Item white noise; S3.2.3, Geological prediction: Based on Network and The model's geological hazard risk assessment model evaluates seismic activity and geological deformation. The calculation process is as follows: ， in, Indicates Item geological risk score, Indicates based on Network and Model of geological disaster risk assessment, Indicates geological deformation data, Indicates Items of seismic activity data; S3.2.

4. Comprehensive flood risk forecast: After smoothing each item Respectively , and Combined into new data , using a Long-term and short-term networks The algorithm predicts the risk rate; use Building a subsequence temporal representation learner ,pass Learning compressed and contextual subsequence representations from long time series, Use mask reconstruction task, and then pass random mask data 75% of the data is masked ,pass right Perform reconstruction optimization to obtain the mask subsequence , and by reconstructing the loss function right For optimization, the calculation process is as follows: ， From the data The remaining 25% of the data is not masked Reconstruct the unmasked subsequence in the temporal context of , and by reconstructing the loss function right For optimization, the calculation process is as follows: ， Masked subsequence and unmasked subsequence Together they constitute the time representation of the processed subsequence ; (1) Long-term trend extraction: From the subsequence time representation To capture long-term trend features, stacked one-dimensional dilated convolutional layers are used to expand the temporal receptive field and capture long-term trends. Layer, the one-dimensional dilated convolution operation is as follows: ， in, represents the output of the first layer of the one-dimensional dilated convolutional layer, Represents subsequence time representation The elements, C represents the convolution kernel of a one-dimensional dilated convolution operation, represents the convolution kernel size, represents the variable in the accumulation operation, represents the expansion rate, Represents a one-dimensional dilated convolution operation; After obtaining the output of the first layer of the one-dimensional dilated convolutional layer, the result of the previous layer is convolved with the convolution kernel of the next layer, and then the output of the next layer is obtained through activation function and maximum pooling operation. The specific operations are as follows: ， …… ， …… ， in, Indicates The output of the layer, , express The operation of the activation function, Represents the maximum pooling, the output of the last layer Recorded as long-term trend characteristics ; (2) Short-term trend extraction: Capturing local fine-grained trend features in short-term historical sequences, using existing spatiotemporal graph neural networks As a short-term trend extractor, to extract short-term trend features, the calculation process is as follows: ， in, Indicates short-term trend characteristics. represents the adjacency matrix; (3) Feature fusion and risk prediction: The long-term trend features and short-term trend features are integrated to obtain the final prediction results. The feature integration and prediction calculation process is as follows: ， in, express of Iteration operation, Represents the prediction result, represents a multi-layer perceptron, Represents a join operation.

2. A reservoir flood risk prediction method according to claim 1, characterized in that: S1 is as follows: Through sensors in reservoir management centers, meteorological monitoring stations, hydrological monitoring stations and geological monitoring stations Interval duration in time Collect hydrological, meteorological and geological data, including reservoir water level , outflow , evapotranspiration , rainfall , meteorological data , earthquake activity data and geological deformation data , meteorological data These include air temperature, humidity and wind speed; Build a dataset based on the collected data , , , , , , dataset Communist Party of China group data, data set The total time of data collection is The time interval for collecting each set of data is , Indicated in The first Group data, , , , , , and Respectively expressed in The first The group includes reservoir water levels, outflow, evapotranspiration, rainfall, meteorological data, seismic activity data and geological deformation data.

3. A reservoir flood risk prediction method according to claim 2, characterized in that S2 The details are as follows: S2.

1. Blockchain network deployment: Treat the reservoir management center, meteorological monitoring station, hydrological monitoring station, and geological monitoring station as separate nodes in the blockchain, and choose private chain as the blockchain network architecture to ensure data security and privacy; S2.

2. Data processing before uploading to the chain: The dataset Each type of data is packaged into data blocks at a certain time interval. Suppose the data collection time is , the time Divide evenly into time intervals, there are data blocks, data block collection Expressed as , Indicates Data blocks, hash each data block, generate a unique hash value to ensure data integrity; Reuse The symmetric encryption algorithm encrypts the data block, and the encryption key is , the specific calculation is as follows: ， in, Indicates that the encryption key is Symmetric encryption algorithm, Represents a collection of data blocks The data in the data block is symmetrically encrypted; S2.

3. Data upload and consensus mechanism: The symmetrically encrypted data The hash value of each data block is uploaded to the nodes in the blockchain network in turn, and each node verifies and confirms the uploaded data through the consensus mechanism of the blockchain.

4. A reservoir flood risk prediction method according to claim 3, characterized in that: S3 is as follows: The data processing center receives the symmetrically encrypted data After calculating the hash value of each data block, the received data is first restored to a data set The initial data collected in the process is then preprocessed for the restored data. The data in the data is preprocessed, and then the preprocessed data is analyzed through big data and based on the self-attention mechanism model Long short-term memory network for R The algorithm performs data analysis to obtain comprehensive flood risk prediction results, and finally conducts risk assessment to generate early warning information.

5. A reservoir flood risk prediction system, which executes a reservoir flood risk prediction method as claimed in any one of claims 1 to 4, characterized in that: Includes the following modules: (1) Data acquisition module: Hydrological monitoring equipment: water level meters and flow meters are deployed at different locations of the reservoir to monitor water level and flow changes in real time; Meteorological monitoring equipment: Rain gauges, temperature and humidity sensors, and anemometers are installed around the reservoir and in the upstream area to collect rainfall, temperature, humidity, and wind speed; Geological monitoring equipment: seismometers, inclinometers, and strain gauges are deployed on the reservoir dam and surrounding areas to monitor seismic activity and geological deformation; (2) Data transmission module: transmits the pre-processed data, and uses blockchain encryption technology to enhance data security during the transmission process; (3) Data analysis module: The preprocessed data is transmitted to the data processing center, which stores and analyzes the received data to generate flood risk prediction results and warning information.

Images