Earth-rock dam leakage thermal infrared image prediction method and device

Abstract

The application discloses a kind of earth-rock dam leakage thermal infrared image prediction method and device, belong to earth-rock dam leakage monitoring and prediction technical field.Its method includes: using distributed thermal infrared ground station to continuously collect the thermal infrared image of earth-rock dam leakage hidden danger area;With time series infrared image input SA-E3D-LSTM network model is trained, and the thermal infrared image change situation of future time period is predicted.Its device includes dual-optical infrared thermal imager, photovoltaic power supply module, data transceiver module, fixed frame, steering device, central server and user terminal.A kind of earth-rock dam leakage thermal infrared image prediction method and device provided by the application can realize the thermal infrared image prediction of earth-rock dam leakage, with the advantages of anti-interference, wide coverage, good space-time continuity, etc., solve the problem that traditional point type leakage prediction method cannot reflect the distribution and development trend of leakage in space.

Description

Technical Field

[0001] This invention relates to the field of seepage monitoring and prediction technology for earth-rock dams, and in particular to a method and device for predicting seepage using thermal infrared images of earth-rock dams. Background Technology

[0002] Earth-rock dams, as crucial water conservancy infrastructure, directly impact the normal operation of flood control, irrigation, and power generation systems through their structural safety. Leakage is a major cause of dam failure; therefore, effectively monitoring and predicting dam leakage is a critical issue in engineering safety management. Failure to detect and address leakage in a timely manner will compromise the structural integrity of the dam, reduce its disaster resistance, threaten its daily operation and maintenance, and may even trigger breaches, dam collapses, and other severe disasters, causing irreversible social losses.

[0003] Seepage in earth-rock dams is characterized by high spatiotemporal randomness and strong concealment. Timely detection and monitoring of seepage development are crucial for ensuring the safety of earth-rock dams. Currently, seepage prediction methods for earth-rock dams are mainly based on point monitoring data such as water level and water pressure, and the prediction results cannot reflect the spatial distribution and development trend of seepage. Summary of the Invention

[0004] The purpose of this invention is to provide a method and device for predicting seepage in earth-rock dams using thermal infrared images. By extracting the spatiotemporal features of thermal infrared image sequences and modeling the time series, the seepage trend of earth-rock dams can be predicted, thereby providing technical and equipment support for seepage prevention design and emergency response to seepage.

[0005] To achieve the above objectives, the present invention provides a method and apparatus for predicting seepage in earth-rock dams using thermal infrared images, which utilizes distributed thermal infrared ground stations to continuously collect thermal infrared images of areas with potential seepage hazards in earth-rock dams.

[0006] The time-series thermal infrared images are input into the SA-E3D-LSTM network model for training, and the model outputs future thermal infrared image sequences.

[0007] The method includes the following steps:

[0008] S1. Data Acquisition: Real-time monitoring of seepage hazard areas of earth-rock embankments is conducted using distributed thermal infrared ground stations. Continuous thermal infrared and visible light image sequences are acquired. The thermal infrared image sequences reflect the temporal and spatial variation trend of surface heat distribution in the seepage hazard area of ​​the earth-rock embankment. The image sequences include heat distribution information on the surface of the earth-rock embankment. Before acquiring thermal infrared and visible light images of the earth-rock embankment at the site, the area of ​​the seepage hazard area is divided, and several distributed thermal infrared ground stations are deployed.

[0009] S2. Data Input and Preprocessing: The input data consists of time-series thermal infrared images. The thermal infrared image sequence data comprises T frames, each with dimensions of H×W×C (height, width, channels). The data is divided into small 3D data blocks, typically within fixed time windows, each window consisting of several frames, forming an H×W×D×C tensor. These 3D data blocks will become the input to the SA-E3D-LSTM network model, where D represents the number of frames (depth dimension) within the time window.

[0010] The input data is represented as follows:

[0011] X t ∈R H×W×C fort = 1, 2, ..., T

[0012] Where: t is the time step, H is the frame height, W is the frame width, and C is the number of channels (e.g., 3 channels for RGB).

[0013] S3. Spatiotemporal feature extraction: Spatiotemporal features are extracted from the preprocessed thermal infrared image sequence through 3D convolution. 3D convolution can extract local spatiotemporal features in both spatial and temporal dimensions.

[0014] S4. Time series modeling: The spatiotemporal features extracted from 3D convolution are modeled in time series using a Long Short-Term Memory (LSTM) network to capture long-term temporal dependencies in thermal infrared image sequences.

[0015] S5. Improved EideticMemory mechanism: During the LSTM hidden state update process, 3D convolution operation is used to reconstruct the spatial features of the cell state, which enhances the ability to remember spatiotemporal features. A spatial attention mechanism is also introduced to weightedly filter the thermal imaging abnormal features of the seepage area, thereby improving the ability to capture the seepage features of the thermal infrared image of the earth-rock dam.

[0016] S6. Thermal Infrared Image Prediction: Based on the output of the LSTM network, predict the changing trend of thermal infrared images of potential seepage areas in earth-rock dams over future time periods, and predict the seepage situation of earth-rock dams.

[0017] S7. Output Results: Output the predicted thermal infrared image and generate a field prediction report related to leakage risk for assessing the safety status of earth-rock dams.

[0018] Preferably, in step S1, the distributed thermal infrared ground station has the ability to stably acquire thermal infrared images and visible light image sequence data in complex environments (such as mountainous, windy, and rainy areas). Compared with UAV flight platforms, the distributed infrared ground station can acquire thermal distribution information on the surface of earth and rock dams more accurately and comprehensively.

[0019] Preferably, in step S3, when the 3D convolution extracts the spatiotemporal features of the thermal infrared image, the convolution kernel slides simultaneously in the spatial and temporal dimensions to capture the features of the thermal distribution on the surface of the earth-rock dam changing over time.

[0020] At each time step t, the input thermal infrared image sequence fragment X t The data is fed into the SA-E3D-LSTM network model. The main function of LSTM is to handle dependencies in the temporal dimension, capturing dynamic information between the current time step and past time steps. The LSTM structure uses three gating mechanisms to update and control the hidden state and the memory state:

[0021] Input gate: Determines the input information to be added to the current memory.

[0022] Forget Gate: Decides which information to discard from current memory.

[0023] Output gate: Determines the output at the current time step.

[0024] For each time step, the mathematical expression of LSTM is as follows:

[0025] i t =σ(W i [h t-1 ,X t ]+b i (Input Gate);

[0026] f t =σ(W f [h t-1 ,X t ]+b f (The Gate of Oblivion);

[0027] o t =σ(W o [h t-1 ,X t ]+b o (Output gate);

[0028] (Candidate memory units);

[0029] (Memory status update);

[0030] h t =o t ⊙tanh(C t (Hidden state);

[0031] Where: i t f t o tThese are the input gate, the forget gate, and the output gate. It is a candidate memory, C t σ is the hidden state, σ is the sigmoid activation function, tanh is the hyperbolic tangent function, and ⊙ is element-wise multiplication.

[0032] LSTM can process the temporal correlation in thermal infrared image sequences, capturing the relationship between each frame and its preceding and following frames.

[0033] Preferably, in step S4, the input to the LSTM network is the spatiotemporal feature map output by 3D convolution, and the LSTM network performs time series dependency modeling on the input spatiotemporal features through a gating mechanism;

[0034] The SA-E3D-LSTM network model introduces 3D convolutional layers to process the spatial dimension of thermal infrared image sequence data. 3D convolution performs convolution not only in space (height and width) but also in the temporal dimension (depth), thus capturing the spatial features that change over time in thermal infrared image sequence data. The mathematical expression for the 3D convolution operation is:

[0035] X out =W c *X in +b c

[0036] Where: * represents a 3D convolution operation, W c It is a 3D convolution kernel, typically a kernel of size kH×kW×kD, where X represents the convolution size in height, width, and depth, respectively. in It is the input data block, X out It is the feature map of the output, b c It is a bias.

[0037] 3D convolution enables models to learn feature changes in input thermal infrared image sequences not only in the spatial dimension (H and W) but also in the temporal dimension (D), thereby capturing complex spatiotemporal relationships.

[0038] Preferably, in step S5, the improved Eidetic memory mechanism adds a spatial attention mechanism during the LSTM hidden state update process, adds a spatial attention module after the 3D convolution output, weights and filters the thermal imaging abnormal features of the seepage area, suppresses background noise (such as solar reflection, vegetation cover, etc.), and makes the model focus on seepage-related thermal infrared features (such as linear temperature anomaly bands), thereby enhancing the model's ability to capture the spatiotemporal features of thermal infrared images of seepage in earth and rock dams.

[0039] The spatial attention mechanism employs a dual-path weighting approach, enabling the model to dynamically focus on areas of abnormal temperature on the surface of the earth-rock dam (such as gradient change zones around seepage points). Channel attention suppresses irrelevant feature channels (such as background vegetation thermal radiation), while spatial attention reinforces localized high-temperature gradient change zones caused by seepage; both work synergistically to enhance the recognizability of seepage features. The attention weight matrix A∈R H×W The mathematical expression is as follows:

[0040] M c =σ(W2·ReLU(W1·AvgPool(X) t ()))(Channel Attention Path)

[0041] M s =σ(Conv 5×5 ([MaxPool(X t AvgPool(X) t Spatial attention path

[0042] A t =λ·M c (X t )+(1-λ)·M s (X t (Attention weight fusion)

[0043] Where W1∈R C / r×C W2∈R C×C / r r = 4 is the channel compression ratio, and σ is the Sigmoid function; Conv 5×5 λ represents a 5×5 convolution kernel, and [;] represents channel concatenation; the learnable parameter λ∈[0,1].

[0044] Preferably, similar to the standard LSTM, the SA-E3D-LSTM network model also updates the cell state to capture long-term dependencies. The updated cell state not only considers the LSTM gating mechanism but also includes an improved EideticMemory mechanism to enhance the hidden state. The update formula is as follows:

[0045] C′ t =A t (Conv3D([h t-1 C t-1 ]))⊙C t

[0046] In the formula: Conv3D is a 3×3×3 3D convolution kernel used to extract spatiotemporal correlation features, and ⊙ is element-wise multiplication.

[0047] This update mechanism combines temporal and spatial information, enhancing memory effectiveness and spatial relevance through an improved EideticMemory.

[0048] Preferably, in time series processing, the SA-E3D-LSTM network model processes thermal infrared image sequence data at multiple time steps. The input at each time step undergoes LSTM and 3D convolution processing to update cell and hidden states. After the entire sequence is completed, the model generates future thermal infrared image frame sequences by analyzing the outputs of all time steps, predicting potential surface temperature changes caused by seepage in earth-rock dams and identifying potential seepage areas.

[0049] Preferably, based on the abnormal changes in the surface temperature of the earth-rock dam in the predicted thermal infrared image sequence, the leakage risk level is assessed, and leakage early warning information is output to help engineering managers take protective measures in advance.

[0050] A thermal infrared image prediction device for seepage in earth-rock dams includes several distributed thermal infrared ground stations, a central server, and several user terminals. The distributed thermal infrared ground stations are deployed in the seepage hazard area of ​​the earth-rock dam, the central server is located in the central control room, and the user terminals are carried by staff. The distributed thermal infrared ground stations are communicatively connected to the central server and the user terminals.

[0051] Preferably, distributed thermal infrared ground stations are deployed in areas with potential seepage risks in earth-rock embankments, specifically including:

[0052] Dual-light infrared imaging module: continuously acquires thermal infrared and visible light images of the potential hazard area of ​​earth-rock embankment at a set sampling frequency;

[0053] Photovoltaic power supply module: including photovoltaic charging panel and battery, used to provide the required power to various modules and devices of the distributed thermal infrared ground station.

[0054] Data transceiver module: includes a data repeater and a wireless transceiver unit. The data transceiver unit is used to transmit the thermal infrared image time series data temporarily stored in the data repeater to the central server.

[0055] The dual-light infrared thermal imager, photovoltaic power supply module, and data transceiver module are all mounted on a fixed frame. The dual-light infrared thermal imager is connected to the fixed frame via a steering device to adjust the shooting angle of the dual-light infrared thermal imager.

[0056] Preferably, the central server includes a central processor, a data preprocessing unit, a data storage unit, and a wireless transceiver unit, used to preprocess and store thermal infrared images and visible light images collected by the distributed thermal infrared ground station, predict thermal infrared images of potential seepage areas in earth-rock dams, and push on-site information to user terminals.

[0057] The information pushed on-site includes the push time, on-site prediction report, thermal infrared and visible light images collected by the distributed thermal infrared ground station at the push time, and the latitude and longitude coordinates of the distributed thermal infrared ground station.

[0058] The user terminal includes a data communication module and a human-computer interaction module, which are used to receive and view thermal infrared image prediction information of seepage hazard areas in earth-rock dams.

[0059] Therefore, the present invention, which employs the above-described structure, provides a method and apparatus for predicting thermal infrared images of seepage in earth-rock dams, and has the following beneficial effects:

[0060] (1) This invention uses distributed thermal infrared ground stations to continuously collect thermal infrared images of seepage hazard areas of earth-rock dams and predict the changes in thermal infrared images in future time periods. It can effectively capture the temperature changes with time and space during the seepage process of earth-rock dams, solve the problem that traditional point-type seepage prediction methods cannot reflect the distribution and development trend of seepage in space, and provide strong technical and equipment support for seepage prevention layout design and emergency response to seepage.

[0061] (2) The data acquisition of this invention has spatiotemporal continuity and is less affected by the environment, and can realize all-weather thermal infrared image data acquisition of seepage hazard areas of earth and rock dams in complex environments.

[0062] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of the system layout of the present invention;

[0064] Figure 2 This is a schematic diagram of a localized low-temperature anomaly in an area without vegetation cover within the seepage hazard zone of an earth-rock dam in Embodiment 1 of the present invention.

[0065] Figure 3 This is a schematic diagram of a localized low-temperature anomaly in an area without vegetation cover within the seepage hazard zone of an earth-rock dam in Embodiment 1 of the present invention.

[0066] Figure 4 This is a schematic diagram of a localized low-temperature anomaly in a vegetation-covered area within a potential seepage zone of an earth-rock dam in Embodiment 1 of the present invention.

[0067] Figure 5 This is a schematic diagram of a localized high-temperature anomaly in the vegetation-covered area within the seepage hazard zone of an earth-rock dam in Embodiment 1 of the present invention.

[0068] Figure 6 This is a schematic diagram of the SA-E3D-LSTM network structure of the method for predicting thermal infrared images of seepage in earth-rock dams according to the present invention.

[0069] Figure 7 The results of infrared image prediction of seepage thermal infiltration in earth-rock dams provided in Embodiment 1 of the present invention;

[0070] Figure 8 This is an evaluation of the infrared image prediction results of seepage thermal infiltration of earth-rock dams provided in Embodiment 1 of the present invention and a comparison with other models.

[0071] Figure 9 This is a schematic diagram of the infrared image prediction device for seepage in earth and rock dams according to the present invention;

[0072] Figure Labels

[0073] 1. Dual-light infrared thermal imager; 2. Photovoltaic power supply module; 3. Data transceiver module; 4. Mounting frame; 5. Steering device; 6. Central server; 7. User terminal. Detailed Implementation

[0074] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0075] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0076] Example 1

[0077] like Figure 1 As shown, the present invention provides a method for predicting thermal infrared images of seepage in earth-rock dams, comprising the following steps:

[0078] Step S1: Divide the potential leakage areas of the earth-rock embankment and select areas to deploy distributed thermal infrared ground stations.

[0079] Step S2: Debug the distributed thermal infrared ground station so that the focal plane of the dual-light thermal imager 1 is parallel to the slope and acquire thermal infrared and visible light image sequences of the seepage hazard area at a frequency of 1 frame / min.

[0080] Step S3: Transmit the thermal infrared images and visible light images collected by the distributed thermal infrared ground station to the central server 6.

[0081] In step S4, the data preprocessing unit in the central server 6 stores the received thermal infrared image and visible light image data in the data storage unit in a one-to-one correspondence format, that is, the two corresponding thermal infrared images and visible light images share the same number, such as 0001-A and 0001-B; and performs preprocessing on the thermal infrared image sequence, including noise reduction, normalization processing and frame resampling of the time series of thermal infrared images.

[0082] Step S5: Input the preprocessed thermal infrared image sequence into the SA-E3D-LSTM network model for model training, output the future thermal infrared image frame sequence, predict the surface temperature change that may be caused by seepage in the earth-rock dam, and identify potential seepage areas.

[0083] Step S6: Based on the abnormal changes in the surface temperature of the earth-rock dam in the predicted thermal infrared image sequence, assess its risk level and output leakage early warning information.

[0084] Step S7: The information push unit of the central server 6 pushes on-site information of the seepage hazard area of ​​the earth-rock dam to the user terminal 7.

[0085] On-site information includes push time, on-site report, and latitude and longitude coordinates of the distributed thermal infrared ground station;

[0086] The on-site report assesses the potential risk level of dam leakage based on changes in temperature distribution in a predicted thermal infrared image sequence.

[0087] In step S8, the engineering management personnel will design the anti-seepage layout and organize emergency response work based on the on-site information of the potential seepage area of ​​the earth-rock embankment on the user terminal 7.

[0088] like Figure 2-5 As shown, the thermal infrared images of the seepage hazard area of ​​the earth-rock dam exhibit temperature variations, including normal, locally low-temperature anomalies, and locally high-temperature anomalies. Low-temperature anomalies and high-temperature anomalies are relative; low-temperature anomalies indicate that the temperature at the seepage outlet is lower than the background temperature (i.e., in non-seepage areas), while high-temperature anomalies indicate that the temperature at the seepage outlet is higher than the background temperature. Thermal infrared images are mapped to color spaces (such as grayscale, RGB, HSV, etc.) based on their temperature values, with the color gradations in the image representing their corresponding temperature values. The contour and texture features of the thermal infrared images of seepage in earth-rock dams are key information for the SA-E3D-LSTM network model to predict thermal infrared images.

[0089] The dataset contains 400 time-series thermal infrared images of areas with potential seepage hazards in earth-rock dams. Before training, the training and validation sets were divided in an 8:2 ratio. The Timestep was set to 10, with one thermal infrared image output per Timestep. The initial learning rate was set to 0.0001, the Batchsize to 32, the maximum number of iterations to 500, and the validation frequency to 50.

[0090] like Figure 6 , Figure 7 As shown, time-series thermal infrared images are input into the SA-E3D-LSTM network model. After training, the model outputs a sequence of future thermal infrared image frames. It can be seen that the predicted frames are very similar to the ground truth in terms of thermal image feature details and color relationships, with only minor differences in image details. Based on the predicted frames, the future development of leakage can be intuitively predicted.

[0091] The MSE and SSIM metrics were used to quantitatively evaluate the model's prediction performance, measuring the similarity between the predicted frame and the ground truth from both the error value and perceptual perspectives. The prediction results of the E3D-LSTM model were calculated and summarized. Figure 8 As can be seen, compared with the E3D-LSTM model, the SA-E3D-LSTM (Spatial-Attention Eidetic 3D-LSTM) model provided by this invention has a higher SSIM value and a lower and more stable MSE value, indicating that the model has a significant improvement in the prediction accuracy of thermal infrared images of seepage in earth-rock dams, and is relatively less affected by background noise.

[0092] like Figure 9 As shown, a thermal infrared image prediction device for seepage in earth and rock dams is used to perform the method described in any of the above-mentioned methods. It includes a plurality of distributed thermal infrared ground stations, a central server 6 and a plurality of user terminals 7, wherein the distributed thermal infrared ground stations are communicatively connected to the central server 6 and the user terminals 7.

[0093] The distributed thermal infrared ground station is deployed in the seepage hazard area of ​​the earth-rock embankment. It includes a dual-light infrared thermal imager 1, a photovoltaic power supply module 2, and a data transceiver module 3. The dual-light infrared thermal imager 1 is electrically connected to the photovoltaic power supply module 2 and the data transceiver module 3 to continuously acquire the thermal infrared image and visible light image time series data of the earth-rock embankment hazard area and send them to the central server 6.

[0094] The dual-light infrared thermal imager 1, photovoltaic power supply module 2, and data transceiver module 3 are all mounted on the fixed frame 4; the dual-light infrared thermal imager 1 is connected to the fixed frame 4 through the turning device 5, which is used to adjust the shooting angle of the dual-light infrared thermal imager 1.

[0095] The main parameters of the dual-light infrared thermal imager 1 used in this embodiment of the invention are shown in Table 1.

[0096] Table 1 Main parameters of Dual-Light Infrared Thermal Imager 1

[0097]

[0098]

[0099] The central server 6 is located in the central control room and includes a central processor, a data preprocessing unit, a data storage unit and a wireless transceiver unit. It is used to preprocess and store thermal infrared time series data uploaded by distributed thermal infrared ground stations, predict thermal infrared images for future time periods and push on-site information of seepage hazard areas of earth and rock embankments to user terminals 7.

[0100] User terminals 7, including a data communication module and a human-computer interaction module, are provided to relevant engineering management personnel to receive and view on-site information on potential seepage areas in earth-rock embankments.

[0101] The information pushed on-site includes the push time, on-site forecast report, thermal infrared and visible light images collected by the distributed thermal infrared ground station at the push time, and the latitude and longitude coordinates of the distributed thermal infrared ground station.

[0102] The equipment components involved in this invention include a dual-light infrared thermal imager, a photovoltaic charging panel, a storage battery, a data repeater, a wireless transceiver unit, a central processing unit, a data preprocessing unit, a data storage unit, an information push unit, and a user terminal. Many product models of the above components can be used in the prior art. Those skilled in the art can select appropriate models according to actual needs. This embodiment will not list them all.

[0103] Therefore, the present invention employs the above-mentioned method and device for predicting seepage in earth-rock dams using thermal infrared images, which can realize the prediction of seepage in earth-rock dams using thermal infrared images. It has the advantages of anti-interference, wide coverage, and good spatiotemporal continuity, and solves the problem that traditional point-based seepage prediction methods cannot reflect the distribution and development trend of seepage in space.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for predicting seepage in earth-rock dams using thermal infrared images, characterized in that: The method comprises the following steps: S1, data acquisition: real-time monitoring of the seepage hidden danger area of the earth and rockfill dam by a distributed thermal infrared ground station, collecting continuous thermal infrared image sequences and visible light image sequences, the thermal infrared image sequences including thermal distribution information of the surface of the earth and rockfill dam; before collecting the thermal infrared images and visible light images of the earth and rockfill dam site, the range of the seepage hidden danger area of the earth and rockfill dam is divided, and a plurality of distributed thermal infrared ground stations are arranged; S2, data preprocessing: preprocessing the thermal infrared image sequences, including noise removal, brightness correction of the thermal infrared images, and time frame resampling of the thermal infrared image sequences, to ensure the consistency and accuracy of the input data; S3, spatio-temporal feature extraction: spatio-temporal feature extraction of the preprocessed thermal infrared image sequences by 3D convolution, which can extract local spatio-temporal features in spatial and temporal dimensions at the same time; S4, time series modeling: time series modeling of the spatio-temporal features extracted from the 3D convolution by the long short-term memory network LSTM, capturing the long-term dependence in time in the thermal infrared image sequences; wherein the long short-term memory network LSTM adopts an improved Eidetic Memory mechanism to enhance the hidden state: in the LSTM hidden state updating process, the cell state is reconstructed by 3D convolution operation to enhance the memory ability of the spatio-temporal features, and a double-path attention mechanism is introduced, the channel attention suppresses irrelevant feature channels, and the spatial attention strengthens the local high temperature gradient change area caused by seepage, both of which cooperate to improve the capture ability of the thermal infrared image seepage features of the earth and rockfill dam; S5, thermal infrared image prediction: based on the output of the LSTM network, predicting the change trend of the thermal infrared image of the earth and rockfill dam seepage hidden danger area in the future time period, and predicting the seepage situation of the earth and rockfill dam; S6, result output: outputting the predicted thermal infrared image and generating a site prediction report related to the seepage risk, for evaluating the safety condition of the earth and rockfill dam.

2. The method according to claim 1, wherein: In step S1, the distributed thermal infrared ground station has the ability to stably collect thermal infrared image and visible light image sequence data in complex environments, and compared with the unmanned aerial vehicle flight platform, the distributed infrared ground station can more accurately and comprehensively obtain the thermal distribution information of the surface of the earth and rockfill dam.

3. The method according to claim 1, wherein: In step S3, when extracting the spatio-temporal features of the thermal infrared image, the convolution kernel slides in spatial and temporal dimensions at the same time, capturing the features of the change of the surface thermal distribution of the earth and rockfill dam over time.

4. The method of claim 1, wherein the method further comprises: In step S4, the input of the LSTM network is the spatio-temporal feature map output by the 3D convolution, and the LSTM network models the time series dependence of the input spatio-temporal features through the gating mechanism.

5. The earth and rockfill dam seepage thermal infrared image prediction method according to claim 1, characterized in that: In step S4, the improved EideticMemory mechanism adds a spatial attention mechanism in the LSTM hidden state update process, aiming at the non-uniformity of the heat conduction of the leakage outlet in the earth-rock dam leakage process, dynamically focuses on the temperature anomaly area through the attention weight, and enhances the network's ability to capture the spatial and temporal dependence in the thermal infrared image sequence; Specifically, the long short-term memory network LSTM adopts an improved EideticMemory mechanism to enhance the hidden state, including: performing double-path attention weighting on the extracted spatio-temporal features; fusing the weighted features with the original cell state to update the cell state.

6. The method of claim 1, wherein the method further comprises: In step S5, the potential leakage area and temperature anomaly on the surface of the earth-rock dam are detected by generating the thermal infrared image change in the future period of time.

7. The method of claim 1, wherein the method further comprises: The input of the thermal infrared image sequence is a plurality of continuous frames, and the time interval of the plurality of continuous frames is adjusted according to the actual monitoring requirement, so as to adapt to the monitoring frequency requirement in different leakage detection scenarios.

8. The method of claim 1, wherein the method further comprises: In step S6, the on-site prediction report includes the thermal infrared image prediction result, the historical thermal infrared image and the visible light image, and based on the change of the temperature distribution in the predicted thermal infrared image sequence, the potential risk level of the dam leakage is evaluated.

9. A device for predicting thermal infrared images of earth-rock dam seepage, characterized in that: A method for performing the method for predicting thermal infrared images of earth-rock dam leakage according to any one of claims 1-8, comprising a plurality of distributed thermal infrared ground stations, a central server and a plurality of user terminals, the distributed thermal infrared ground stations are arranged in the earth-rock dam leakage hidden danger area, the central server is arranged in the central control room, and the user terminal is carried by the staff, and the distributed thermal infrared ground station is in communication connection with the central server and the user terminal.

10. The device for predicting thermal infrared images of seepage of earth-rock dams according to claim 9, characterized in that: The distributed thermal infrared ground station comprises a dual-optical infrared thermal imager, a photovoltaic power supply module and a data transceiver module, the dual-optical infrared thermal imager is electrically connected with the photovoltaic power supply module and the data transceiver module, and is used for continuously acquiring thermal infrared images and visible light images of the earth-rock dam hidden danger area and sending them to the central server; The dual-optical infrared thermal imager, the photovoltaic power supply module and the data transceiver module are all installed on the fixed frame, and the dual-optical infrared thermal imager is connected with the fixed frame through the steering device, which is used for adjusting the shooting angle of the dual-optical infrared thermal imager; The central server comprises a central processor, a data preprocessing unit, a data storage unit and a wireless transceiver unit, which is used for preprocessing and storing the thermal infrared images and visible light images collected by the distributed thermal infrared ground station, predicting the thermal infrared images of the earth-rock dam on-site leakage hidden danger area and pushing the on-site information to the user terminal; The on-site push information includes push time, on-site prediction report, thermal infrared images and visible light images collected by the distributed thermal infrared ground station at the push moment, and the longitude and latitude coordinates of the distributed thermal infrared ground station; The user terminal comprises a data communication module and a man-machine interaction module, which is used for receiving and viewing the thermal infrared image prediction information of the earth-rock dam leakage hidden danger area.

Images