Water conservancy project safety early warning method, device and equipment based on BIM (Building Information Modeling) and medium

Abstract

The invention relates to a BIM-based water conservancy project safety early warning method and device, equipment and a medium. The method comprises the following steps: acquiring multi-source data such as an osmotic pressure gradient field based on a predefined hydraulic engineering monitoring point space coordinate list, and generating a three-dimensional osmotic field matrix after noise reduction, standardization and interpolation processing; physical characteristics of the matrix are extracted by using a pre-trained space-time convolution model, and abnormal parameter labels are determined through threshold comparison and deviation calculation; associating the exception to a BIM component node through a spatial topology matching algorithm; and finally, performing visual early warning based on the BIM abnormal coordinate vector, and generating a structured alarm instruction set by combining risk levels and abnormal type matching treatment measures. According to the method, through multi-source data fusion modeling, feature analysis and extraction, BIM space mapping, three-dimensional visualization decision making and the like, the precision, efficiency and intuition of hydraulic engineering safety early warning are improved, and a reliable technical scheme is provided for guaranteeing safe and stable operation of hydraulic engineering facilities.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of water conservancy project operation management, and particularly relates to a water conservancy project safety early warning method, device, equipment and medium based on BIM. BACKGROUND

[0002] Water conservancy project safety early warning is a key link to ensure the stable operation of water conservancy facilities and prevent major accidents such as dam collapse. However, the traditional water conservancy project safety early warning method mainly relies on manual inspection and single-point sensor monitoring, which has many shortcomings. On the one hand, the monitoring data obtained by the sensor is usually discrete point information, which is difficult to fully reflect the seepage condition of the water conservancy project, and the data processing process is relatively cumbersome, which cannot efficiently extract effective physical characteristic parameters, thereby affecting the timely discovery and accurate judgment of safety hazards. On the other hand, this method usually uses a single threshold overrun for alarm processing, lacks correlation analysis capability, and is prone to false alarms or missed alarms. In addition, although Building Information Modeling (BIM) is widely used in water conservancy project design, it is only used as a static geometric display tool in the traditional water conservancy project safety early warning method. Moreover, the monitoring data and BIM components lack spatial topological correlation, such as the inability to match abnormal parameter label sets to BIM component node IDs, the need for manual annotation of abnormal positions on two-dimensional drawings, and the difficulty in intuitively displaying three-dimensional risk characteristics such as crack propagation direction. SUMMARY

[0003] Therefore, it is necessary to provide a water conservancy project safety early warning method, device, equipment and medium based on BIM to improve the reliability and scientificity of water conservancy project safety early warning, realize accurate and efficient safety early warning of water conservancy projects, and ensure the safe and stable operation of water conservancy facilities.

[0004] In a first aspect, the present application provides a water conservancy project safety early warning method based on BIM, comprising:

[0005] Based on a pre-defined water conservancy project monitoring point spatial coordinate list, seepage parameter data is collected to generate a multi-source data set with spatial position labels, the seepage parameter data including seepage pressure gradient field data, humidity diffusion field data and crack trajectory sequence data, and the multi-source data set with spatial position labels is processed through noise reduction, standardization transformation and spatial interpolation to generate a standardized three-dimensional seepage field matrix;

[0006] Physical characteristic parameters are extracted from the standardized three-dimensional seepage field matrix through a pre-trained spatio-temporal convolution model to generate a seepage parameter quantization set, and based on the seepage parameter quantization set and a pre-set standard parameter library, an abnormal parameter label set is generated through threshold comparison and deviation calculation, the abnormal parameter label set including abnormal types and deviation rates;

[0007] Based on the abnormal parameter label set and the water conservancy BIM model, the BIM component node matching processing is performed through the spatial topology matching algorithm, and the BIM abnormal coordinate vector is generated;

[0008] According to the BIM abnormal coordinate vector, the risk thermal coloring processing and the crack direction marking processing are performed, the three-dimensional visual early warning interface is generated, and based on the three-dimensional visual early warning interface, the risk level identification and the disposal measure matching processing are performed, the structured alarm instruction set is generated, and the structured alarm instruction set includes the BIM component node ID, the risk level and the disposal measure.

[0009] In one of the embodiments, the standardized three-dimensional seepage field matrix is subjected to physical characteristic parameter extraction through the pre-trained spatiotemporal convolution model, and a seepage parameter quantization set is generated, including:

[0010] The pre-trained spatiotemporal convolution model is called, wherein the spatiotemporal convolution model includes an input layer, a spatial convolution layer, a time derivative layer, an acceleration feature layer and an output layer, and the input layer dimension of the spatiotemporal convolution model and the data dimension of the standardized three-dimensional seepage field matrix are matched with each other;

[0011] The standardized three-dimensional seepage field matrix is input into the spatiotemporal convolution model, and the seepage pressure gradient modulus is calculated through the spatial convolution layer to obtain the seepage pressure gradient modulus;

[0012] The humidity diffusion rate is calculated and processed through the time derivative layer to obtain the humidity diffusion rate;

[0013] The crack propagation acceleration containing the direction vector is obtained through the acceleration feature layer for crack propagation acceleration calculation;

[0014] The seepage pressure gradient modulus, the humidity diffusion rate and the crack propagation acceleration are respectively bound to the spatial coordinates of the standardized three-dimensional seepage field matrix through spatial coordinate association processing to generate the seepage parameter quantization set.

[0015] In one of the embodiments, based on the seepage parameter quantization set and the pre-set standard parameter library, the abnormal parameter label set is generated through threshold comparison and deviation calculation, and the abnormal parameter label set includes the abnormal type and the deviation rate, including:

[0016] The seepage pressure gradient threshold range, the humidity diffusion rate threshold range and the crack acceleration threshold range are called from the pre-set standard parameter library;

[0017] Based on the seepage pressure gradient threshold range, the humidity diffusion rate threshold range and the crack acceleration threshold range, the absolute difference value of each parameter value in the seepage parameter quantization set from the corresponding threshold range median value is calculated to generate the deviation value;

[0018] When the deviation value exceeds the preset deviation threshold of the corresponding threshold range, the corresponding parameter value is marked as an abnormal type based on the preset abnormal classification rule to obtain the abnormal type. The deviation rate is calculated using the deviation rate formula, and the abnormal parameter label set is generated in combination with the spatial coordinates corresponding to the parameter value.

[0019] In one embodiment, based on the abnormal parameter label set and the water conservancy project BIM model, a BIM component node matching process is performed using a spatial topology matching algorithm to generate a BIM abnormal coordinate vector, including:

[0020] Based on the BIM model of the water conservancy project, a topological relationship map including the spatial positions of component nodes is generated through a topological relationship extraction algorithm;

[0021] Through the spatial coordinates in the abnormal parameter label set, the nearest neighbor search algorithm is used to match the BIM component nodes in the topological relationship graph to obtain the nearest BIM component node ID;

[0022] The abnormality type and deviation rate in the abnormal parameter label set are associated with the corresponding nearest BIM component node ID through attribute binding processing to generate a BIM abnormality coordinate vector.

[0023] In one embodiment, risk thermal coloring and crack direction marking are performed based on BIM abnormal coordinate vectors to generate a 3D visual warning interface. Based on the 3D visual warning interface, risk level identification and treatment measure matching are performed to generate a structured alarm instruction set. The structured alarm instruction set includes the BIM component node ID, risk level, and treatment measures, including:

[0024] According to the deviation rate in the BIM abnormal coordinate vector, the risk level is converted through a linear mapping function to generate a risk thermal value corresponding to the BIM component node ID in the BIM abnormal coordinate vector;

[0025] Based on the hydraulic engineering BIM model, color rendering is performed on the component nodes corresponding to the risk thermal values. For component nodes with abnormal cracks, the direction vector of the crack propagation acceleration is marked based on the quantified seepage parameter, generating a 3D visual warning interface.

[0026] Based on the 3D visual warning interface, the BIM component node ID is extracted and processed and the risk level is identified through the graphic analysis engine to obtain the risk level;

[0027] Based on the risk level and the anomaly type in the BIM anomaly coordinate vector, the decision rule engine matches the operation instructions in the preset disposal strategy library and generates disposal measures;

[0028] Combined with BIM component node ID, risk level and disposal measures, a structured alarm instruction set is generated.

[0029] In one embodiment, based on a preset list of spatial coordinates of water conservancy project monitoring points, seepage parameter data is collected to generate a multi-source data set with spatial location tags, including:

[0030] Based on the safety analysis report of the key structure of the water conservancy project, a predefined list of three-dimensional spatial coordinates of monitoring points is obtained, wherein the coordinates of each monitoring point in the predefined three-dimensional spatial coordinate list are pre-associated with the corresponding BIM component node ID during the design phase;

[0031] According to the predefined three-dimensional spatial coordinate list of monitoring points, the sensor network is used to collect the seepage gradient field data, the humidity diffusion field data and the crack trajectory sequence data;

[0032] The seepage pressure gradient field data, humidity diffusion field data and crack trajectory sequence data are respectively bound to the corresponding monitoring point spatial coordinates and acquisition timestamps to generate a multi-source data set with spatial location tags.

[0033] In one embodiment, a standardized three-dimensional seepage field matrix is ​​generated by performing noise reduction, normalization transformation, and spatial interpolation on a multi-source dataset with spatial position markers, including:

[0034] Based on the multi-source dataset with spatial position markers, the environmental noise is filtered through the wavelet transform algorithm to generate a denoised multi-source dataset;

[0035] The Z-score normalization algorithm is used to normalize the denoised multi-source data set to generate a standardized numerical matrix;

[0036] According to the standardized numerical matrix, the three-dimensional spatial field is continuously reconstructed through the Kriging spatial interpolation algorithm to generate a standardized three-dimensional seepage field matrix.

[0037] Secondly, the present application also provides a BIM-based water conservancy project safety early warning device, including:

[0038] The data acquisition and preprocessing module is used to collect seepage parameter data based on a predefined list of spatial coordinates of water conservancy project monitoring points, and generate a multi-source data set with spatial location tags. The seepage parameter data includes seepage pressure gradient field data, humidity diffusion field data, and crack trajectory sequence data. The multi-source data set with spatial location tags is processed through noise reduction, normalization transformation, and spatial interpolation to generate a standardized three-dimensional seepage field matrix.

[0039] The feature analysis and anomaly identification module is configured to extract physical feature parameters from the standardized three-dimensional seepage field matrix through a pre-trained spatio-temporal convolution model, generate a seepage parameter quantization set, and generate an abnormal parameter label set based on the seepage parameter quantization set and a pre-set standard parameter library through threshold comparison and deviation calculation, the abnormal parameter label set including an abnormal type and a deviation rate;

[0040] The component matching and coordinate mapping module is configured to perform BIM component node matching processing through a spatial topology matching algorithm based on the abnormal parameter label set and the water conservancy engineering BIM model, and generate a BIM abnormal coordinate vector.

[0041] The early warning analysis and risk disposal module is configured to perform risk heat coloring processing and crack direction marking processing based on the BIM abnormal coordinate vector, generate a three-dimensional visual early warning interface, and perform risk level identification and disposal measure matching processing based on the three-dimensional visual early warning interface, and generate a structured alarm instruction set, the structured alarm instruction set including a BIM component node ID, a risk level and a disposal measure.

[0042] In a third aspect, the present application also provides a computer device, including a memory and a processor, the memory stores a computer program, and the processor implements the steps in the first aspect when executing the computer program.

[0043] In a fourth aspect, the present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps in the first aspect.

[0044] The above-mentioned BIM-based water conservancy engineering safety early warning method, device, equipment and medium provide a data basis for comprehensively mastering the seepage state of the water conservancy engineering by collecting seepage parameter data based on pre-defined monitoring point coordinates and generating a multi-source data set with spatial position markers, and constructing a standardized three-dimensional seepage field matrix based on the pre-processed data set. Secondly, the three-dimensional seepage field matrix is subjected to feature extraction, and a label set including an abnormal type and a deviation rate is generated in combination with a standard parameter library, thereby further improving the accuracy of anomaly identification and early warning capability. Moreover, the abnormal parameter label set is mapped with the BIM model component node to generate a BIM abnormal coordinate vector, which can realize deep integration of monitoring data and engineering structure and provide a direct basis for early warning risk positioning. In addition, risk heat coloring and crack direction marking are performed based on the BIM abnormal coordinate vector to generate a three-dimensional visual early warning interface, and a structured alarm instruction set is generated through risk level identification and disposal measure matching, which can convert abstract data into intuitive spatial risk display and standardized disposal scheme, thereby significantly improving decision-making efficiency and emergency response speed and providing automatic and intelligent decision support for water conservancy facility operation and maintenance.

[0045] Compared with the traditional water conservancy engineering safety early warning method, the method effectively solves the problems of poor spatial continuity of monitoring data, low abnormality recognition accuracy, and non-intuitive risk positioning, and improves the accuracy, efficiency and intuitiveness of water conservancy engineering safety early warning, and provides a more efficient and reliable technical solution for ensuring the safe and stable operation of water conservancy facilities. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 A flow chart of a BIM-based water conservancy engineering safety early warning method is provided for an exemplary embodiment of the present application.

[0048] Figure 2 A structural schematic diagram of a BIM-based water conservancy engineering safety early warning device is provided for an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the present application more clear and obvious, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0050] In one embodiment, as shown in Figure 1 , a BIM-based water conservancy engineering safety early warning method is provided, and the present embodiment takes the method applied to a terminal as an example. It should be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and can be realized through the interaction of the terminal and the server. In the present embodiment, the method includes the following steps:

[0051] S101: Based on the pre-defined water conservancy engineering monitoring point space coordinate list, collecting seepage parameter data, generating a multi-source data set with spatial position markers, the seepage parameter data including seepage pressure gradient field data, humidity diffusion field data and crack trajectory sequence data, and through noise reduction processing, standardization transformation and spatial interpolation processing, a standardized three-dimensional seepage field matrix is generated.

[0052] Specifically, according to the pre-defined water conservancy project monitoring point spatial coordinate list, various sensors such as osmotic pressure meter, humidity sensor, crack monitor, etc. can be arranged on the water conservancy project site. Based on the sensors, by setting the sampling frequency, the osmotic pressure gradient field data, humidity diffusion field data and crack trajectory sequence data can be continuously collected. Among them, the osmotic pressure gradient field data reflects the change of seepage pressure at different positions inside the water conservancy project structure, the humidity diffusion field data reflects the diffusion and distribution of water in the structure material, and the crack trajectory sequence data records the development trajectory of the crack over time. And when collecting data, the spatial position information corresponding to each data point can be marked, so as to generate a multi-source data set with spatial position marking. In addition, the collected original data may be disturbed by environmental noise, sensor error and other factors. Therefore, noise interference in the data can be removed through noise reduction processing to improve the data quality. Through standardization transformation, data of different dimensions and different ranges are converted into a unified standard form for subsequent processing. And the data gaps between monitoring points can also be filled through spatial interpolation processing, and then a complete three-dimensional seepage field matrix is generated, providing a data basis for subsequent seepage analysis. The finally generated standardized three-dimensional seepage field matrix comprehensively and accurately describes the seepage field state of the water conservancy project.

[0053] S102: physical feature parameter extraction of the standardized three-dimensional seepage field matrix is performed through the pre-trained spatio-temporal convolution model to generate a seepage parameter quantization set, and based on the seepage parameter quantization set and the pre-set standard parameter library, an abnormal parameter label set is generated through threshold comparison and deviation calculation, the abnormal parameter label set including abnormal type and deviation rate.

[0054] Specifically, the pre-trained spatio-temporal convolution model is obtained by training on a large amount of historical water conservancy project seepage data. The model combines convolution operations in time and space dimensions, and can effectively extract physical feature parameters in the standardized three-dimensional seepage field matrix to form a seepage parameter quantization set. The pre-set standard parameter library stores the standard range and reference value of various seepage parameters of the water conservancy project in the normal operating state. By comparing the seepage parameter quantization set with the data in the standard parameter library, whether the parameter exceeds the normal range can be judged through threshold comparison, and it can be considered as abnormal. Finally, the deviation calculation is performed to generate an abnormal parameter label set. The label set includes abnormal type and deviation rate. Through the label set, the category and severity of the anomaly can be determined, providing a scientific basis for subsequent risk assessment.

[0055] S103: based on the abnormal parameter label set and the water conservancy project BIM model, BIM component node matching processing is performed through a spatial topology matching algorithm to generate a BIM abnormal coordinate vector.

[0056] Specifically, the BIM model of the water conservancy project is a three-dimensional digital modeling of the entire water conservancy project, which contains detailed information and spatial topological relations of each component of the project. Based on the model, the matching component nodes in the BIM model can be searched based on the spatial position information corresponding to the abnormal parameters in the abnormal parameter label set through the spatial topological matching algorithm, and then the positioning of the abnormal position is realized, and the BIM abnormal coordinate vector is generated. The vector records the three-dimensional coordinate information of the specific BIM component node where the anomaly is located, providing accurate position identification for subsequent visualization and risk analysis.

[0057] S104: According to the BIM abnormal coordinate vector, risk heat coloring processing and crack direction marking processing are performed, a three-dimensional visualization early warning interface is generated, and based on the three-dimensional visualization early warning interface, risk level identification and disposal measure matching processing are performed, and a structured alarm instruction set is generated, which includes BIM component node ID, risk level and disposal measure.

[0058] Specifically, according to the BIM abnormal coordinate vector generated in S103, different colors and shades can be used to perform risk heat coloring processing on the abnormal position in the BIM model, for example, red represents a high-risk area, yellow represents a medium-risk area, and green represents a low-risk area. Moreover, if the BIM abnormal coordinate vector has a crack anomaly, the crack direction marking processing can be performed in the BIM model according to the crack direction information in the vector, so as to further present the development direction and morphology of the crack.

[0059] The crack data is marked with the crack direction in the BIM model to further present the development direction and morphology of the crack. Through the above processing, a three-dimensional visualization early warning interface can be finally generated, and the operation and maintenance personnel can intuitively understand the abnormal situation and risk distribution of the water conservancy project through the interface. In addition, based on the interface, the deviation rate, abnormal type and historical risk data of the abnormal parameters can be combined to identify the risk level according to the preset rules, and the risk can be divided into three levels of low, medium and high. Then, according to the identified risk level and abnormal type, the corresponding disposal measures are matched from the disposal measure library to generate a structured alarm instruction set. The instruction set contains key information such as BIM component node ID, risk level and disposal measure, providing a clear emergency response guide for the operation and maintenance personnel, ensuring that measures can be taken in time and effectively to eliminate safety hazards and further ensure the safe operation of the water conservancy project.

[0060] In the above method, the seepage parameter data is collected by the pre-defined monitoring point spatial coordinate list and a multi-source data set with spatial position markers is generated, and then noise reduction, standardization and spatial interpolation processing are performed to generate a standardized three-dimensional seepage field matrix. This can convert discrete monitoring data into continuous three-dimensional spatial distribution model data, and then comprehensively and intuitively present the internal seepage state of the water conservancy project, providing a data basis for subsequent anomaly detection. Secondly, the pre-trained spatio-temporal convolution model is used to extract features from the standardized three-dimensional seepage field matrix, and combined with the seepage parameter quantization set and the pre-set standard parameter library to generate an abnormal parameter label set, realizing the analysis and abnormal identification of the seepage field multi-parameter, further improving the efficiency and accuracy of the safety hazard detection. Further, based on the abnormal parameter label set and the water conservancy BIM model, the BIM abnormal coordinate vector is generated by the spatial topological matching algorithm, realizing the direct mapping from data anomaly to component risk, providing the basis for accurately positioning the problem part. In addition, according to the BIM abnormal coordinate vector, the risk thermal coloring and crack direction marking are performed to generate a three-dimensional visual warning interface, and the risk level identification and disposal measure matching are performed to generate a structured alarm instruction set, so that the management personnel can intuitively master the risk distribution and development trend, realizing the standardized and automated output of the warning information, and providing a powerful guarantee for the intelligent operation and maintenance of water conservancy facilities.

[0061] For the multi-source data set with spatial position markers, a standardized three-dimensional seepage field matrix is generated through noise reduction processing, standardization transformation and spatial interpolation processing, including:

[0062] Based on the multi-source data set with spatial position markers, a denoised multi-source data set is generated by wavelet transform algorithm for environmental noise filtering processing;

[0063] The Z-score standardization algorithm is used to normalize the denoised multi-source data set to generate a standardized numerical matrix;

[0064] According to the standardized numerical matrix, a standardized three-dimensional seepage field matrix is generated by Kriging spatial interpolation algorithm for three-dimensional spatial field continuous reconstruction processing.

[0065] Specifically, based on the multi-source data set with spatial position markers, the multi-source data set with spatial position markers can be decomposed according to the monitoring points to obtain the time series signal of each monitoring point. Then, the Daubechiesdb4 wavelet basis function can be used to perform multi-layer wavelet decomposition on each time series signal to decompose the signal into subbands of different frequencies. In the wavelet domain, the soft threshold denoising method can be used to set the coefficients less than the threshold to zero and shrink the coefficients greater than the threshold.

[0066] Finally, the signal is reconstructed by inverse wavelet transform to effectively remove power frequency interference, sensor white noise and other environmental noise, and retain the mutation characteristics of the seepage parameter signal, to obtain the denoised time series. After denoising the signals of all monitoring points, the denoised multi-source data set can be generated by recombination. Then, the Z-score standardization algorithm can be used to normalize the denoised multi-source data set, convert the data to a standard normal distribution with a mean of 0 and a standard deviation of 1, and generate a standardized numerical matrix.

[0067] The Kriging spatial interpolation algorithm is an interpolation method based on spatial autocorrelation, which can estimate the data of unknown points according to the data of known points. Illustratively, the variogram between each data point in the standardized numerical matrix can be calculated first. The variogram reflects the spatial correlation between data points, and is usually fitted using an exponential model, a Gaussian model or a spherical model. According to the fitting results of the variogram, the parameters of Kriging interpolation can be determined, such as the base value and the range of spatial correlation. And in the interpolation process, considering the anisotropic characteristics of water conservancy structures, different range parameters can be set in different directions. For example, the horizontal range can be set to twice the vertical range to reflect the layered characteristics of structures such as dams. Subsequently, for each grid point, the weight coefficient can be calculated according to the spatial position of the surrounding monitoring points and the variogram model, and the parameter estimation value of the node can be obtained by weighted average, and the estimation variance can be calculated to evaluate the interpolation accuracy. Finally, the Kriging interpolation algorithm can optimize the interpolation results by minimizing the estimation variance, generating a three-dimensional seepage field matrix with high spatial continuity and accuracy. This matrix completely describes the three-dimensional spatial distribution and temporal evolution characteristics of the seepage field of water conservancy projects, providing basic data for subsequent safety analysis.

[0068] In one embodiment, the standardized three-dimensional seepage field matrix is subjected to physical feature parameter extraction by a pre-trained spatio-temporal convolution model to generate a seepage parameter quantization set, including:

[0069] The pre-trained spatio-temporal convolution model is called, wherein the spatio-temporal convolution model includes an input layer, a spatial convolution layer, a time derivative layer, an acceleration feature layer and an output layer, and the input layer dimension of the spatio-temporal convolution model matches the data dimension of the standardized three-dimensional seepage field matrix;

[0070] The standardized three-dimensional seepage field matrix is input into the spatio-temporal convolution model, and the spatial convolution layer is used to calculate the seepage pressure gradient modulus to obtain the seepage pressure gradient modulus;

[0071] The time derivative layer is used for humidity diffusion rate calculation and processing to obtain the humidity diffusion rate;

[0072] The acceleration feature layer is used for crack propagation acceleration calculation to obtain crack propagation acceleration containing a direction vector;

[0073] The seepage pressure gradient modulus, humidity diffusion rate and crack propagation acceleration are respectively bound to the spatial coordinates of the standardized three-dimensional seepage field matrix through spatial coordinate association processing to generate a quantitative set of seepage parameters.

[0074] Specifically, the spatiotemporal convolutional model can be built based on a residual network architecture and trained using the Adam optimizer. Its loss function can be a combination of mean squared error loss and L1 regularization to balance the model's fitting ability and generalization performance. The model's input layer can adopt a three-dimensional tensor structure, with its dimensions set to the spatial coordinate dimension, the time series dimension, and the number of percolation parameter channels, aligned with the data dimensions of the standardized three-dimensional percolation field matrix.

[0075] After inputting the standardized three-dimensional seepage field matrix into the spatiotemporal convolution model, the osmotic pressure gradient modulus can be calculated through the spatial convolution layer to obtain the osmotic pressure gradient modulus. This spatial convolution layer can use a multi-scale convolution kernel design, such as 3×3×3 and 5×5×5 convolution kernels, to extract spatial features at different scales through parallel convolution operations. Specifically, a spatial difference operation can first be performed on the input standardized three-dimensional seepage field matrix to calculate the osmotic pressure change rate at each spatial point in the X, Y, and Z coordinate directions, forming a three-dimensional osmotic pressure gradient vector field. Subsequently, the osmotic pressure gradient modulus is calculated through vector modulus calculation. The time derivative layer can adopt a gated recurrent unit structure, which can effectively capture the time series characteristics of humidity data. Schematically, the input standardized three-dimensional seepage field matrix can be expanded along the time dimension to form a spatial humidity distribution matrix. Subsequently, for each spatial point, the humidity change between adjacent time steps is calculated and divided by the time interval to obtain the humidity diffusion rate at that point.

[0076] Furthermore, the acceleration feature layer can employ a spatiotemporal attention mechanism to focus on key areas and moments of crack propagation. Specifically, the fracture trajectory sequence data in the input standardized three-dimensional seepage field matrix can be preprocessed to extract the spatial coordinate sequence of the fracture tip. Subsequently, the acceleration components of the fracture tip in each direction are calculated using the second-order central difference method. By combining the acceleration components in each direction into a three-dimensional acceleration vector, the modulus of this vector can represent the magnitude of the fracture propagation acceleration, and the vector direction represents the direction of the fracture propagation acceleration. Ultimately, the acceleration feature layer can output a fracture propagation acceleration feature tensor of dimension [X, Y, Z, 3], i.e., the fracture propagation acceleration, where each spatial point corresponds to a three-dimensional acceleration vector.

[0077] Specifically, a spatial coordinate mapping relationship between the feature map and the original seepage field matrix can be established first, and then the eigenvalue of each target spatial point can be calculated by a bilinear interpolation method. Finally, the seepage pressure gradient modulus, humidity diffusion rate and crack propagation acceleration can be spliced according to the channel dimension to form the seepage parameter quantization set.

[0078] In one embodiment, based on the seepage parameter quantization set and the preset standard parameter library, an abnormal parameter label set is generated through threshold comparison and deviation calculation, the abnormal parameter label set including an abnormal type and a deviation rate, including:

[0079] The seepage pressure gradient threshold range, humidity diffusion rate threshold range and crack acceleration threshold range are called from the preset standard parameter library.

[0080] Based on the seepage pressure gradient threshold range, humidity diffusion rate threshold range and crack acceleration threshold range, the absolute difference between each parameter value in the seepage parameter quantization set and the median value in the corresponding threshold range is calculated to generate a deviation value.

[0081] When the deviation value exceeds the preset deviation threshold of the corresponding threshold range, the corresponding parameter value is marked with an abnormal type based on a preset abnormal classification rule, the abnormal type is obtained, the deviation rate is calculated by a deviation rate formula, and the abnormal parameter label set is generated in combination with the spatial coordinates corresponding to the parameter value.

[0082] Specifically, the preset standard parameter library can include three sub-libraries of design standard library, historical database and expert knowledge base. The design standard library is used to store the safety threshold of each parameter specified in the water conservancy engineering design file, the historical database can store the historical monitoring data and statistical analysis results in the engineering operation process, and the expert knowledge base can store the preset parameter threshold adjustment rule. Illustratively, when the threshold range is called, the basic threshold range can be obtained from the design standard library according to the current engineering operation state, then the threshold range is dynamically corrected through the historical database, and finally the threshold range is secondarily calibrated by using the expert knowledge base to obtain the seepage pressure gradient threshold range, humidity diffusion rate threshold range and crack acceleration threshold range. Based on the above threshold range, the absolute difference between each parameter value in the seepage parameter quantization set and the median value in the corresponding threshold range can be calculated to generate a deviation value.

[0083] When the deviation value exceeds the preset deviation threshold corresponding to the threshold range, for example, exceeds 15% of the osmotic pressure gradient threshold range, 20% of the humidity diffusion rate threshold range, and 10% of the crack acceleration threshold range, a preset anomaly classification rule based on a decision tree structure is adopted, the corresponding parameter value is processed for anomaly type labeling according to the degree of exceeding the threshold and the parameter type, the anomaly type is obtained, the deviation rate is calculated through the deviation rate formula, and the abnormal parameter label set is generated in combination with the spatial coordinates corresponding to the parameter value. And in order to more accurately quantify the degree of anomaly, the deviation rate can also be calculated through the deviation rate formula. The deviation rate is a relative index, which can reflect the proportion of the degree of deviation of the parameter value from the normal range and the size of the normal range. Finally, in combination with the spatial coordinates corresponding to the parameter value, that is, the spatial coordinates, anomaly type and deviation rate of each abnormal parameter are associated and stored, forming a triple structure including spatial coordinates, anomaly type and deviation rate. Further, by sorting all the triple sets of abnormal parameters, an abnormal parameter label set matrix can be generated.

[0084] In one embodiment, based on the abnormal parameter label set and the water conservancy engineering BIM model, BIM component node matching processing is performed through a spatial topology matching algorithm to generate a BIM abnormal coordinate vector, including:

[0085] Based on the water conservancy engineering BIM model, a topology relationship graph containing component node spatial positions is generated through a topology relationship extraction algorithm;

[0086] Through the spatial coordinates in the abnormal parameter label set, BIM component node matching processing is performed in the topology relationship graph through a nearest neighbor search algorithm to obtain the nearest BIM component node ID;

[0087] The anomaly type and the deviation rate in the abnormal parameter label set are associated to the corresponding nearest BIM component node ID through attribute binding processing to generate a BIM abnormal coordinate vector.

[0088] Specifically, the water conservancy project BIM model can be stored in IFC (Industry Foundation Classes) standard data format, containing three-dimensional geometric information, component attribute information and spatial topological relationship. The topological relationship extraction algorithm can first perform semantic analysis on the BIM model, and identify main structural components such as dam body, gate, water conveying pipeline and corresponding spatial boundaries. Then the boundary representation method is used to extract the geometric elements such as vertices, edges and faces of each component. Further, in the topological relationship graph construction process, a graph database can be used to store component nodes and their relationships. Each component node contains information such as unique identifier (ID), three-dimensional spatial coordinates, component type and material properties. The relationship between nodes can include spatial adjacency relationship, structural connection relationship and support relationship, etc. Finally, a directed graph containing multiple component nodes and multiple relationship edges, i.e. the topological relationship graph, can be obtained.

[0089] Illustratively, each abnormal point in the abnormal parameter label set contains three-dimensional spatial coordinates, which can be mapped to a specific component node in the BIM model using a nearest neighbor search algorithm based on KD tree. Specifically, the coordinates of all component nodes in the topological relationship graph can be constructed into a KD tree index. For each abnormal point coordinate, a fast search can be performed in the KD tree to find the nearest component node in terms of Euclidean distance, i.e. the nearest neighbor BIM component node ID. Further, a maximum search radius threshold can also be set, and when the distance of the nearest neighbor node exceeds the threshold, it can be determined as a non-matching node and processed again. In the secondary processing, a spatial interpolation method can be used to calculate the most likely matching node by weighted average according to the positions and properties of multiple component nodes around the abnormal point. After matching, a mapping relationship table between the abnormal point and the BIM component node can be established to record the nearest neighbor BIM component node ID corresponding to each abnormal point. Then, the abnormal type and deviation rate in the abnormal parameter label set are associated to the corresponding nearest neighbor BIM component node ID through attribute binding processing to generate a BIM abnormal coordinate vector. Finally, the abnormal coordinate vectors of all matched BIM component nodes are combined to obtain a BIM abnormal coordinate vector matrix. The matrix can accurately reflect the spatial distribution and severity of abnormal conditions in the water conservancy project.

[0090] In one embodiment, according to the BIM abnormal coordinate vector, risk thermodynamic coloring processing and crack direction marking processing are performed to generate a three-dimensional visualization warning interface, and based on the three-dimensional visualization warning interface, risk level identification and disposal measure matching processing are performed to generate a structured warning instruction set, which includes BIM component node ID, risk level and disposal measure, including:

[0091] According to the deviation rate in the BIM abnormal coordinate vector, risk level conversion is performed through a linear mapping function to generate a risk thermal value corresponding to the BIM component node ID in the BIM abnormal coordinate vector;

[0092] Based on the water conservancy BIM model, color rendering processing is performed on the component nodes corresponding to the risk thermal value, and for the component nodes with crack abnormality, crack direction marking processing is performed based on the direction vector of the concentrated crack propagation acceleration quantified by the seepage parameter to generate a three-dimensional visual warning interface.

[0093] Based on the three-dimensional visual warning interface, BIM component node ID extraction processing and risk level identification processing are performed through a graphic analysis engine to obtain the risk level.

[0094] Based on the risk level and the abnormal type in the BIM abnormal coordinate vector, operation instruction matching processing is performed in the preset disposal strategy library through a decision rule engine to generate disposal measures.

[0095] Combined with the BIM component node ID, the risk level and the disposal measures, a structured alarm instruction set is generated.

[0096] Specifically, the linear mapping function can be constructed by piecewise linear transformation, that is, the deviation rate can be mapped to the risk thermal value [0, 1] interval, for example, the higher the thermal value, the higher the risk level. Subsequently, based on the water conservancy BIM model, the HSV color model can be used to perform color rendering processing on the component nodes corresponding to the risk thermal value, and the color can be selected according to the thermal value, such as using green for low thermal value and red for high thermal value, so that the risk distribution can be intuitively displayed. And an arrow symbol can be used for crack direction marking, that is, for each component node with crack abnormality, a three-dimensional line segment with an arrow can be drawn at the corresponding position of the BIM model according to the direction vector of the concentrated crack propagation acceleration quantified by the seepage parameter. The arrow length is proportional to the modulus of the crack propagation acceleration, and the acceleration value is mapped to the preset arrow length range through linear mapping. And the arrow color can be gradually changed from light red to dark red according to the acceleration size, to enhance the visual differentiation degree. Finally, WebGL technology can be used to generate a three-dimensional visual warning interface to support real-time interaction and multi-angle browsing.

[0097] Based on the three-dimensional visualization early warning interface, the BIM component node ID corresponding to each pixel is determined by the graphic analysis engine first, and then the HSV color value is restored to the risk heat value through color space conversion. The risk level can be obtained through the preset level rule. In addition, based on the risk level and the abnormal type in the BIM abnormal coordinate vector, the operation instruction matching processing can be carried out in the preset disposal strategy library through the decision rule engine to generate disposal measures. The disposal measures can be sorted by priority, and each measure contains information such as execution priority, measure code, detailed description, responsible person and completion deadline. Finally, combined with the BIM component node ID, the risk level and the disposal measures, a structured alarm instruction set can be generated. The structured alarm instruction set can be directly sent to the water conservancy project management system to support automatic execution and manual intervention processing methods.

[0098] In one embodiment, based on the preset water conservancy monitoring point spatial coordinate list, the seepage parameter data is collected to generate a multi-source data set with spatial position markers, including:

[0099] Based on the water conservancy key structure safety analysis report, a pre-defined monitoring point three-dimensional spatial coordinate list is obtained, wherein each monitoring point coordinate in the pre-defined monitoring point three-dimensional spatial coordinate list is pre-associated with a corresponding BIM component node ID in the design stage;

[0100] According to the pre-defined monitoring point three-dimensional spatial coordinate list, the seepage pressure gradient field data, humidity diffusion field data and crack trajectory sequence data are collected through the preset sensor network;

[0101] The seepage pressure gradient field data, humidity diffusion field data and crack trajectory sequence data are respectively bound with the corresponding monitoring point spatial coordinates and collection time stamps to generate a multi-source data set with spatial position markers.

[0102] Specifically, based on the pre-stored safety analysis report of the key structure of the water conservancy project, a pre-defined monitoring point three-dimensional spatial coordinate list can be obtained, wherein each monitoring point coordinate in the pre-defined monitoring point three-dimensional spatial coordinate list is pre-associated with a corresponding BIM component node ID at the design stage. The safety analysis report of the key structure of the water conservancy project is the basis for the arrangement of the monitoring points. The monitoring point three-dimensional spatial coordinate list can include fields such as monitoring point number, X coordinate, Y coordinate, Z coordinate, and associated BIM component node ID. In the design stage, through the parametric design function of the BIM model, a corresponding parametric family can be constructed for each monitoring point and positioned to the corresponding structure position. Each monitoring point family contains a unique identifier, which establishes a mapping relationship with the BIM component node ID and is stored in the attribute database of the BIM model. The preset sensor network can adopt a distributed heterogeneous sensor system architecture. Among them, the seepage pressure gradient field data can be monitored in real time through the fiber grating sensor array, the humidity diffusion field data can be detected and collected by the distributed wireless humidity sensor network, and the crack trajectory sequence data can be collected by the camera arranged in the visible area such as the dam surface and the gate chamber.

[0103] The seepage pressure gradient field data, the humidity diffusion field data, and the crack trajectory sequence data are bound to the corresponding monitoring point spatial coordinates and the collection time stamps, respectively, so that a multi-source data set with spatial position markers can be generated.

[0104] The multi-source data set can be stored in a hierarchical data format, including a seepage pressure data set, a humidity data set, and a crack data set, and each main data set contains spatial coordinates, time stamps, and measurement values. The association between different data sources can be established through an indexing mechanism to support time and space-based queries.

[0105] As shown in Figure 2 Based on the same inventive concept, the embodiments of the present application also provide a BIM-based water conservancy project safety early warning device 200 for implementing the above-mentioned related methods. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more BIM-based water conservancy project safety early warning device embodiments provided below can be referred to the limitations of the BIM-based water conservancy project safety early warning method in the above, which will not be repeated here. The device includes:

[0106] The data acquisition and preprocessing module 201 is configured to acquire seepage parameter data based on the pre-defined water conservancy monitoring point spatial coordinate list, generate a multi-source data set with spatial position markers, the seepage parameter data including seepage pressure gradient field data, humidity diffusion field data, and crack trajectory sequence data, and generate a standardized three-dimensional seepage field matrix through noise reduction processing, standardization transformation, and spatial interpolation processing on the multi-source data set with spatial position markers.

[0107] The feature analysis and anomaly identification module 202 is configured to extract physical feature parameters from the standardized three-dimensional seepage field matrix through a pre-trained spatio-temporal convolution model, generate a seepage parameter quantization set, and based on the seepage parameter quantization set and a pre-set standard parameter library, generate an abnormal parameter label set through threshold comparison and deviation calculation, the abnormal parameter label set including an abnormal type and a deviation rate.

[0108] The component matching and coordinate mapping module 203 is configured to perform BIM component node matching processing through a spatial topology matching algorithm based on the abnormal parameter label set and the water conservancy engineering BIM model, and generate a BIM abnormal coordinate vector.

[0109] The early warning analysis and risk disposal module 204 is configured to perform risk heat coloring processing and crack direction marking processing according to the BIM abnormal coordinate vector, generate a three-dimensional visual early warning interface, and based on the three-dimensional visual early warning interface, perform risk level identification and disposal measure matching processing, and generate a structured alarm instruction set, the structured alarm instruction set including a BIM component node ID, a risk level and a disposal measure.

[0110] In the above device, the data acquisition and preprocessing module 201 acquires seepage parameter data through a pre-defined monitoring point spatial coordinate list and generates a multi-source data set with spatial position markers, and then generates a standardized three-dimensional seepage field matrix after preprocessing, which can fully integrate discrete seepage data and provide a reliable data basis for subsequent seepage feature analysis. The feature analysis and anomaly identification module 202 extracts physical feature parameters using a pre-trained spatio-temporal convolution model to generate a seepage parameter quantization set, and compares it with a pre-set standard parameter library to generate an abnormal parameter label set including an abnormal type and a deviation rate, which can accurately identify seepage abnormalities and further improve the accuracy of early warning compared to a single threshold alarm method. The component matching and coordinate mapping module 203 generates a BIM abnormal coordinate vector through BIM component node matching based on the abnormal parameter label set and the water conservancy engineering BIM model, which can accurately locate the actual abnormal position. The early warning analysis and risk disposal module 204 performs risk heat coloring and crack direction marking according to the BIM abnormal coordinate vector to generate a three-dimensional visual early warning interface, which can intuitively present the risk characteristics, and through risk level identification and disposal measure matching, output the corresponding structured alarm instruction set, providing clear and intuitive risk information and decision support for management personnel.

[0111] In an exemplary embodiment, the present application also provides a computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the BIM-based safety early warning method for water conservancy projects of the present application when executing the computer program. Preferably, a multi-core processor is used to improve the parallel processing capability of the system. The memory provides sufficient temporary storage space to support the running of the program and the processing of data. The memory capacity should be large enough to accommodate a large amount of supply information and computing tasks.

[0112] In an exemplary embodiment, the present application also provides a computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of the BIM-based safety early warning method for water conservancy projects of the present application.

[0113] The above-described embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the patent scope of the application. It should be noted that, for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application.

Claims

1. A BIM-based water conservancy project safety early warning method, characterized in that: The method comprises: Based on a predefined list of spatial coordinates of water conservancy project monitoring points, seepage parameter data are collected to generate a multi-source data set with spatial location tags. The seepage parameter data includes seepage pressure gradient field data, humidity diffusion field data, and crack trajectory sequence data. The multi-source data set with spatial location tags is subjected to noise reduction, normalization transformation, and spatial interpolation processing to generate a standardized three-dimensional seepage field matrix. Extracting physical characteristic parameters of the standardized three-dimensional seepage field matrix through a pre-trained spatiotemporal convolution model to generate a quantized seepage parameter set, and generating an abnormal parameter label set based on the quantized seepage parameter set and a preset standard parameter library through threshold comparison and deviation calculation, wherein the abnormal parameter label set includes an abnormality type and a deviation rate; Based on the abnormal parameter label set and the water conservancy project BIM model, BIM component node matching processing is performed through a spatial topology matching algorithm to generate a BIM abnormal coordinate vector; According to the BIM abnormal coordinate vector, risk thermal coloring processing and crack direction marking processing are performed to generate a three-dimensional visual warning interface. Based on the three-dimensional visual warning interface, risk level identification and disposal measure matching processing are performed to generate a structured alarm instruction set, which includes the BIM component node ID, risk level and disposal measures.

2. The method according to claim 1, characterized in that The physical characteristic parameters of the standardized three-dimensional seepage field matrix are extracted by using a pre-trained spatiotemporal convolution model to generate a quantized set of seepage parameters, including: Calling the pre-trained spatiotemporal convolution model, wherein the spatiotemporal convolution model includes an input layer, a spatial convolution layer, a time derivative layer, an acceleration feature layer, and an output layer, and the input layer dimension of the spatiotemporal convolution model matches the data dimension of the standardized three-dimensional seepage field matrix; Inputting the standardized three-dimensional seepage field matrix into the spatiotemporal convolution model, and calculating the seepage pressure gradient modulus through the spatial convolution layer to obtain the seepage pressure gradient modulus; Performing humidity diffusion rate calculation processing through the time derivative layer to obtain the humidity diffusion rate; Calculating the crack extension acceleration using the acceleration characteristic layer to obtain the crack extension acceleration including the direction vector; The seepage pressure gradient modulus, the moisture diffusion rate and the crack extension acceleration are respectively bound to the spatial coordinates of the standardized three-dimensional seepage field matrix through spatial coordinate association processing to generate the seepage parameter quantization set.

3. The method according to claim 2, characterized in that Based on the percolation parameter quantization set and the preset standard parameter library, an abnormal parameter label set is generated through threshold comparison and deviation calculation. The abnormal parameter label set includes an abnormality type and a deviation rate, including: Recalling the osmotic pressure gradient threshold range, the moisture diffusion rate threshold range, and the crack acceleration threshold range from the preset standard parameter library; Based on the seepage pressure gradient threshold range, the moisture diffusion rate threshold range, and the crack acceleration threshold range, for each parameter value in the seepage parameter quantization set, calculate the absolute difference between the value and the median of the corresponding threshold range to generate a deviation value; When the deviation value exceeds the preset deviation threshold corresponding to the threshold range, the corresponding parameter value is marked as an abnormal type based on the preset abnormal classification rule to obtain the abnormal type, the deviation rate is calculated by the deviation rate formula, and the spatial coordinates corresponding to the parameter value are combined to generate the abnormal parameter label set.

4. The method according to claim 1, wherein The method of performing BIM component node matching processing based on the abnormal parameter label set and the water conservancy project BIM model by using a spatial topology matching algorithm to generate a BIM abnormal coordinate vector includes: Based on the water conservancy project BIM model, a topological relationship map including the spatial positions of component nodes is generated by a topological relationship extraction algorithm; Using the spatial coordinates in the abnormal parameter tag set, a BIM component node matching process is performed in the topological relationship graph using a nearest neighbor search algorithm to obtain the nearest BIM component node ID; The abnormality type and deviation rate in the abnormal parameter label set are associated with the corresponding nearest BIM component node ID through attribute binding processing to generate the BIM abnormality coordinate vector.

5. The method according to claim 2, characterized in that According to the BIM abnormal coordinate vector, risk thermal coloring processing and crack direction marking processing are performed to generate a three-dimensional visual warning interface, and based on the three-dimensional visual warning interface, risk level identification and disposal measure matching processing are performed to generate a structured alarm instruction set. The structured alarm instruction set includes the BIM component node ID, risk level and disposal measures, including: According to the deviation rate in the BIM abnormal coordinate vector, a risk level conversion is performed through a linear mapping function to generate a risk thermal value corresponding to the BIM component node ID in the BIM abnormal coordinate vector; Based on the water conservancy project BIM model, by performing color rendering processing on the component nodes corresponding to the risk thermal values, and for the component nodes where the abnormality type is the presence of crack abnormalities, performing crack direction marking processing based on the direction vector of the crack propagation acceleration in the seepage parameter quantification concentration, the three-dimensional visual warning interface is generated; Based on the three-dimensional visual warning interface, a BIM component node ID extraction process and a risk level identification process are performed through a graphic analysis engine to obtain the risk level; Based on the risk level and the anomaly type in the BIM anomaly coordinate vector, a decision rule engine is used to perform operation instruction matching processing in a preset disposal strategy library to generate the disposal measure; The structured alarm instruction set is generated by combining the BIM component node ID, the risk level and the disposal measures.

6. The method according to claim 1, characterized in that The method of collecting seepage parameter data based on a preset list of spatial coordinates of water conservancy project monitoring points and generating a multi-source data set with spatial location tags includes: Based on the safety analysis report of the key structure of the water conservancy project, obtaining the predefined three-dimensional spatial coordinate list of monitoring points, wherein the coordinates of each monitoring point in the predefined three-dimensional spatial coordinate list of monitoring points are pre-associated with the corresponding BIM component node ID during the design stage; According to the predefined three-dimensional spatial coordinate list of monitoring points, the seepage pressure gradient field data, the humidity diffusion field data and the crack trajectory sequence data are collected through a preset sensor network; The seepage pressure gradient field data, the humidity diffusion field data and the crack trajectory sequence data are respectively bound to the corresponding monitoring point spatial coordinates and acquisition timestamps to generate the multi-source data set with spatial position tags.

7. The method according to claim 1, characterized in that The multi-source data set with spatial position markers is subjected to noise reduction, normalization transformation and spatial interpolation processing to generate a standardized three-dimensional seepage field matrix, including: Based on the multi-source dataset with spatial position markers, environmental noise filtering is performed using a wavelet transform algorithm to generate a noise-reduced multi-source dataset; The Z-score normalization algorithm is used to normalize the noise-reduced multi-source data set to generate a standardized numerical matrix; According to the standardized numerical matrix, a three-dimensional spatial field continuous reconstruction process is performed using a Kriging spatial interpolation algorithm to generate the standardized three-dimensional seepage field matrix.

8. A BIM-based water conservancy project safety early warning device, characterized in that: The device comprises: A data acquisition and preprocessing module is used to collect seepage parameter data based on a predefined list of spatial coordinates of water conservancy project monitoring points, generate a multi-source data set with spatial position tags, and perform noise reduction, normalization transformation and spatial interpolation on the multi-source data set with spatial position tags to generate a standardized three-dimensional seepage field matrix; A feature analysis and anomaly identification module is used to extract physical feature parameters of the standardized three-dimensional seepage field matrix through a pre-trained spatiotemporal convolution model to generate a quantized set of seepage parameters, and based on the quantized set of seepage parameters and a preset standard parameter library, generate an abnormal parameter label set through threshold comparison and deviation calculation, wherein the abnormal parameter label set includes an anomaly type and a deviation rate; A component matching and coordinate mapping module is used to perform BIM component node matching processing based on the abnormal parameter label set and the water conservancy project BIM model through a spatial topology matching algorithm to generate a BIM abnormal coordinate vector; The early warning analysis and risk disposal module is used to perform risk thermal coloring processing and crack direction marking processing according to the BIM abnormal coordinate vector, generate a three-dimensional visual early warning interface, and based on the three-dimensional visual early warning interface, perform risk level identification and disposal measure matching processing to generate a structured alarm instruction set, which includes the BIM component node ID, risk level and disposal measures.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images