Intelligent diagnosis method and system for dam safety state cooperatively driven by mechanism data

Abstract

The invention relates to a mechanism data collaboratively driven dam safety state intelligent diagnosis method and system. The method comprises the steps of collecting environmental quantity data, dam deformation data and corresponding monitoring point coordinates; a PINN model is constructed, the PINN model comprises an input layer, a hidden layer and an output layer, the input layer receives the collected data, and the output layer outputs a predicted value of a deformation component; the PINN model is trained, the weight and bias of the PINN model are adjusted by minimizing a comprehensive loss function, and the comprehensive loss function comprises a data driving loss function, a physical constraint loss function and a boundary condition loss function; using the trained PINN model to predict a deformation field; and comparing the prediction result with the monitoring index to obtain an abnormal region, and performing abnormality diagnosis on the abnormal region. Through the mechanism data synergistic effect, it is guaranteed that model prediction follows the physical law, actual monitoring data can be better fitted, and high-precision dam safety state prediction and diagnosis are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent dams, and in particular to an intelligent diagnosis method and system for dam safety status driven by mechanism data collaboration. Background Art

[0002] A large number of studies have been conducted at home and abroad on dam safety monitoring, and many results have been achieved, such as diagnosis and abnormality judgment of concrete dam crack behavior, diagnosis and hazard analysis of earth-rock dam seepage behavior, and data-driven safety monitoring and prediction and early warning models for monitoring physical quantities such as deformation, seepage, stress and strain. However, the above-mentioned related studies still have problems such as insufficient fusion of multi-source data, weak synergy between data and mechanism, and insufficient self-diagnosis ability in complex environments. For example, methods based on data and machine learning or deep learning algorithms rely solely on monitoring data for training and lack the constraints of physical laws. This not only leads to poor generalization ability of the model and poor performance when facing data that has never appeared, but also its prediction results are difficult to explain from a physical level, which is not conducive to relevant personnel making accurate decisions in actual engineering applications. Although the method based on monitoring data inversion and numerical calculation can analyze the safety status of the dam to a certain extent, it has extremely high requirements on data quality. Moreover, most of the current methods perform parameter inversion according to homogeneous materials. This method has poor adaptability when facing complex working conditions. Once there is noise interference in the data, it is easy to cause inaccurate parameter inversion. In addition, the inversion and numerical simulation calculation process of this method is time-consuming and cannot meet the urgent needs of real-time monitoring and early warning of dams.

[0003] In view of the above problems, there is an urgent need to develop an intelligent diagnosis method for dam safety status driven by mechanism data collaboration, so as to effectively solve the defects of existing methods and improve the accuracy, reliability and real-time performance of dam safety diagnosis. Summary of the invention

[0004] To this end, the technical problem to be solved by the present invention is to overcome the technical defects faced by the self-diagnosis of dam safety status under complex operating conditions in the prior art, such as being greatly affected by noise data, poor generalization ability, poor adaptability to complex working conditions, complex calculation process, difficulty in physical interpretation of prediction results, and inability to monitor and warn in real time, and propose a method and system for intelligent diagnosis of dam safety status driven by mechanism data collaboration.

[0005] In order to solve the above technical problems, the present invention provides a dam safety status intelligent diagnosis method driven by mechanism data collaboration, comprising the following steps: S10: Collect environmental quantity data, dam deformation data and their corresponding monitoring point coordinates; S20: constructing a PINN model, wherein the PINN model includes an input layer, a hidden layer and an output layer, receiving deformation data, monitoring point coordinates and environmental quantity data through the input layer, using the hidden layer to capture nonlinear features of the input data, and outputting the predicted value of the dam deformation component in the output layer; S30: constructing a comprehensive loss function and training the PINN model, adaptively adjusting the weights and biases of the PINN model by minimizing the comprehensive loss function to obtain a trained PINN model, wherein the comprehensive loss function includes a data-driven loss function, a physical constraint loss function, and a boundary condition loss function; S40: Use the trained PINN model to predict the dam deformation field; S50: Compare the prediction result with the monitoring index to identify the abnormal area, perform abnormal diagnosis on the abnormal area, and trigger an early warning according to the abnormal diagnosis result.

[0006] In one embodiment of the present invention, in S20, the hidden layer is a multi-layer fully connected network, and the activation function of the hidden layer is ReLU or tanh.

[0007] In one embodiment of the present invention, in S30, the method for constructing a comprehensive loss function includes: The comprehensive loss function is the weighted sum of the data-driven loss function, the physical constraint loss function, and the boundary condition loss function, which is expressed as: ; In the formula, represents the comprehensive loss function, represents the data-driven loss function, represents the physical constraint loss function, represents the boundary condition loss function, , and They represent the weights of the data-driven loss function, the physical constraint loss function, and the boundary condition loss function, respectively, where: ; ; ; In the formula, Indicates the number of monitoring points involved in the calculation, Indicates The spatial coordinates and time points of the monitoring points, Indicates the PINN model for the The predicted value of the three-dimensional deformation component of each monitoring point, Indicates Measured values ​​of three-dimensional deformation components at each monitoring point; Indicates the number of physical constraint points involved in the calculation, represents the divergence of the stress tensor, represents the gradient operator, represents the stress tensor, represents body force; represents the fixed boundary condition loss function, represents the water pressure boundary condition loss function.

[0008] In one embodiment of the present invention, in S30, the PINN model is trained, and the method of adaptively adjusting the weights and biases of the PINN model by minimizing the comprehensive loss function includes: S31: Select an optimizer to train the PINN model, randomly initialize the PINN model weights, and calculate the data-driven loss function, physical constraint loss function, and boundary condition loss function respectively; S32: Calculate a comprehensive loss function according to the data-driven loss function, the physical constraint loss function and the boundary condition loss function; S33: Use the automatic differentiation tool to calculate the gradient of the comprehensive loss function to the PINN model parameters, and use the optimizer to update the weights and biases of the PINN model. Repeat the operation until the comprehensive loss function converges or reaches the specified number of training times to obtain the optimal parameters of the PINN model.

[0009] In one embodiment of the present invention, in S50, the prediction result is compared with the monitoring index to identify an abnormal area, and the method for abnormal diagnosis of the abnormal area includes: S51: Based on the historical monitoring data, the safety threshold of the deformation distribution and the safety threshold of the deformation rate are determined by using the confidence interval method, and the predicted value of the dam spatial deformation is predicted by using the trained PINN model, and the predicted value of the dam spatial deformation rate is calculated based on the predicted value of the dam spatial deformation; S52: calculating a normalized deformation index according to a safety threshold of deformation distribution and a predicted value of dam spatial deformation, and calculating a normalized rate index according to a safety threshold of deformation rate and a predicted value of dam spatial deformation rate; S53: Determine the dam safety score by weight coefficient method based on the normalized deformation index and the normalized rate index; S54: Trigger an early warning based on the dam safety score and physical residual classification.

[0010] In one embodiment of the present invention, in S52, the calculation formulas of the normalized deformation index and the normalized rate index are respectively: ; ; In the formula, NDI represents the normalized deformation index, represents the predicted value of the dam spatial deformation at a certain monitoring point, represents the safety threshold of deformation distribution, NVI represents the normalized rate index, represents the predicted value of the spatial deformation rate of the dam at a certain monitoring point, Indicates the safety threshold of deformation rate.

[0011] In one embodiment of the present invention, in S53, the calculation formula of the dam safety score is: ; In the formula, S represents the dam safety score, and represents the weight coefficient, NDI represents the normalized deformation index, NVI represents the normalized rate index.

[0012] Based on the same inventive concept, the present invention also provides a dam safety status intelligent diagnosis system driven by mechanism data collaboration, including: A data acquisition module, which is used to collect environmental quantity data, dam deformation data and their corresponding monitoring point coordinates; A model building module is used to build a PINN model, wherein the PINN model includes an input layer, a hidden layer and an output layer, receives deformation data, monitoring point coordinates and environmental quantity data through the input layer, uses the hidden layer to capture nonlinear features of the input data, and outputs the predicted value of the dam deformation component in the output layer; A model training optimization module, which is used to construct a comprehensive loss function and train the PINN model, and adaptively adjust the weights and biases of the PINN model by minimizing the comprehensive loss function to obtain a trained PINN model, wherein the comprehensive loss function includes a data-driven loss function, a physical constraint loss function, and a boundary condition loss function; The deformation prediction module is used to predict the dam deformation field using the trained PINN model; The abnormal diagnosis and early warning module compares the prediction results with the monitoring indicators to identify the abnormal area, performs abnormal diagnosis on the abnormal area, and triggers an early warning based on the abnormal diagnosis results.

[0013] Based on the same inventive concept, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method when executing the program.

[0014] Based on the same inventive concept, the present invention also provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the above-mentioned method are implemented.

[0015] The above technical solution of the present invention has the following advantages compared with the prior art: 1. The present invention provides a method and system for intelligent diagnosis of dam safety status driven by mechanism data synergistically. The comprehensive loss function constructed by the present invention includes a data-driven loss function, a physical constraint loss function and a boundary condition loss function. When training the PINN model, the weight and bias of the model are adaptively adjusted by minimizing the comprehensive loss function to achieve the coordinated optimization of the physical constraint loss and the data-driven loss. Through the synergistic effect of the mechanism data, it can not only ensure that the model prediction follows the physical laws, but also better fit the actual monitoring data, and finally achieve a higher-precision prediction and intelligent diagnosis of the dam safety status. It has the significant advantages of high reliability, strong practicality and convenient technology promotion, which provides an important guarantee for the long-term safety of reservoir dam projects and provides advanced technical support for the digital and intelligent development of water conservancy projects. 2. The present invention provides a mechanism data collaboratively driven dam safety dynamic intelligent diagnosis method and system, which constrains the model output through a physical constraint loss function to strictly satisfy the physical control equation, and constrains the model output through a boundary condition loss function to meet fixed conditions and water pressure conditions, ensuring that the output of the PINN model has physical meaning and engineering applicability, so that the model has strong generalization ability, can adapt to different working conditions, regions and engineering conditions, and can make reliable predictions for unknown areas without relying on large-scale training data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings.

[0017] Figure 1 It is a flow chart of a dam safety status intelligent diagnosis method driven by mechanism data collaboration proposed by the present invention.

[0018] Figure 2 It is a method flow chart of step S30 of the present invention.

[0019] Figure 3 It is a method flow chart of step S50 of the present invention.

[0020] Figure 4 It is a schematic diagram of the training process of the PINN model of the present invention.

[0021] Figure 5It is a module schematic diagram of a dam safety status intelligent diagnosis system driven by mechanism data collaboration of the present invention. DETAILED DESCRIPTION

[0022] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0023] Reference Figure 1 As shown, an embodiment of the present invention provides a dam safety state intelligent diagnosis method driven by mechanism data collaboration, comprising the following steps: S10: Collect environmental quantity data, dam deformation data and their corresponding monitoring point coordinates; S20: constructing a PINN model, which includes an input layer, a hidden layer and an output layer. The input layer receives deformation data, monitoring point coordinates and environmental quantity data, and uses the hidden layer to capture nonlinear features of the input data, and outputs the predicted value of the dam deformation component in the output layer; S30: constructing a comprehensive loss function and training the PINN model, adaptively adjusting the weight and bias of the PINN model by minimizing the comprehensive loss function to obtain a trained PINN model, wherein the comprehensive loss function includes a data-driven loss function, a physical constraint loss function, and a boundary condition loss function; S40: Use the trained PINN model to predict the dam deformation field; S50: Compare the prediction result with the monitoring index to identify the abnormal area, perform abnormal diagnosis on the abnormal area, and trigger an early warning according to the abnormal diagnosis result.

[0024] The present invention provides a mechanism data collaboratively driven dam safety state intelligent diagnosis method, the constructed comprehensive loss function includes a data-driven loss function, a physical constraint loss function and a boundary condition loss function. When the PINN model is trained, the weight and bias of the model are adaptively adjusted by minimizing the comprehensive loss function to achieve collaborative optimization of the physical constraint loss and the data-driven loss. Through the synergistic effect of the mechanism data, it is ensured that the model prediction follows the physical laws and can better fit the actual monitoring data, and finally achieves higher-precision dam safety state prediction and intelligent diagnosis. It has the significant advantages of high reliability, strong practicality and convenient technology promotion, etc. It provides an important guarantee for the long-term safety of reservoir dam projects and provides advanced technical support for the digital and intelligent development of water conservancy projects.

[0025] Among them, in step S10, the dam will be deformed due to various factors during long-term use. The deformation data can intuitively reflect the actual state of the dam and is a key indicator for evaluating the safety of the dam. The deformation data of the dam monitoring point in this embodiment may include information such as displacement, stress and strain. Each monitoring point has its specific coordinates, which clarify the position of the monitoring point in the dam space. Through the coordinate information, the deformation data can be corresponded to the specific position, so as to accurately understand the deformation of different parts of the dam, which helps to locate the abnormal deformation area of ​​the dam and provide accurate location information for subsequent analysis and processing. Environmental factors also have a significant impact on the safety of the dam. Environmental quantity data covers multiple aspects, including temperature, water level, rainfall, etc. By comprehensively considering the deformation data, monitoring point coordinates and environmental quantity data, the actual state of the dam can be fully and accurately grasped. By analyzing these data, the causes and mechanisms of dam deformation can be deeply understood, providing a scientific basis for the safety assessment of the dam.

[0026] Furthermore, after collecting the dam deformation data and its corresponding monitoring point coordinates, as well as environmental quantity data, they can be subjected to data cleaning, data standardization and data fusion processing. Data cleaning mainly removes outliers and missing values ​​in the data; data standardization normalizes the data to a unified and appropriate range (such as [-1,1]); and data fusion integrates real-time monitoring data with historical data to generate model input.

[0027] Among them, in step S20, the constructed PINN model includes an input layer, a hidden layer and an output layer, wherein the input layer is used as the entrance of the neural network model to receive external input data, including deformation data (such as displacement, stress, strain, etc.) of the dam monitoring point, coordinates of the monitoring point and environmental quantity data (such as water level, temperature, rainfall, etc.). These data are a comprehensive description of the dam state and its environment. They carry various information that may be related to the dam deformation and provide a basis for the learning and prediction of subsequent models; the hidden layer is a multi-layer fully connected network, and the activation function uses ReLU or tanh, which is responsible for capturing complex nonlinear features; the output layer outputs the predicted value of the three-dimensional deformation component of the dam. The data enters the PINN neural network model from the input layer. First, the deformation data, monitoring point coordinates and environmental quantity data are combined in the input layer, and these data are passed to the hidden layer in the form of vectors or matrices. In the hidden layer, the data is processed by multiple neurons in turn, and each neuron performs weighted summation on the input data and transforms it through a nonlinear activation function to obtain new feature representations. These new feature representations are continuously transmitted and transformed in the hidden layer, and more complex and representative features are gradually extracted. Finally, the data processed by the hidden layer reaches the output layer. The output layer generates the predicted value of the dam deformation component through linear or nonlinear transformation according to the feature information transmitted from the hidden layer.

[0028] Among them, in step S30, in the dam deformation analysis, the physical information is embedded in the PINN model to determine the physical control equation of the dam deformation. In practical applications, elastic mechanics equations and elastoplastic mechanics equations can be used as control equations, and the physical constraint loss function is defined by the residual of the physical control equation. Here, the elastic mechanics equation is used as an example for explanation: ; ; ; In the formula, represents the gradient operator, represents the stress tensor, D represents the material stiffness matrix, represents the strain tensor, represents external forces (such as water pressure), Represents a displacement vector.

[0029] At the same time, the boundary conditions of the dam should be considered to be fixed constraints and water pressure. The fixed constraint means that the displacement of the bottom of the dam is zero; the water pressure means that the surface of the upstream side of the dam is subjected to the net water pressure. , where represents the water density, represents the acceleration due to gravity, Indicates water depth.

[0030] Based on the above, the constructed comprehensive loss function is composed of the weighted sum of the data-driven loss function, the physical constraint loss function, and the boundary condition loss function. Among them, the data-driven loss function is used to measure the error between the model prediction value and the actual monitoring data, which can be expressed as: ; In the formula, represents the data-driven loss function, Indicates the number of monitoring points involved in the calculation, Indicates The spatial coordinates and time points of the monitoring points, It means that the PINN model The predicted value of the three-dimensional deformation component of each monitoring point, Indicates The measured values ​​of the three-dimensional deformation components of the monitoring points.

[0031] The above physical constraint loss function is the residual of the control equation, which can be expressed as: ; In the formula, is the physical constraint loss function, Indicates the number of physical constraint points involved in the calculation, represents the divergence of the stress tensor, represents the gradient operator, represents the stress tensor, Represents volume force, including external forces such as gravity and water pressure acting on the dam.

[0032] The above boundary condition loss function is the sum of the fixed boundary loss function and the water pressure boundary loss function, which can be expressed as: ; In the formula, represents the boundary condition loss function, represents the fixed boundary condition loss function, represents the water pressure boundary condition loss function.

[0033] Among them, the fixed boundary condition loss function is: ; In the formula, represents the fixed margin loss function, Indicates the number of sampling points involved in the calculation of fixed boundary loss, Indicates The spatial coordinates of the fixed boundary sampling points, Represents the predicted value of the deformation component of the PINN model at a fixed boundary point.

[0034] The water pressure boundary condition loss function is: ; In the formula, represents the water pressure boundary condition loss function, Indicates the number of sampling points involved in the calculation of water pressure boundary loss, represents the stress tensor predicted by the PINN model, represents the unit normal vector of the boundary, which indicates the direction of stress action; represents the hydrostatic pressure, represents the density of water, represents the acceleration due to gravity, Indicates water depth.

[0035] Based on the above, the comprehensive loss function of this embodiment is the weighted sum of the data-driven loss function, the physical constraint loss function and the boundary condition loss function, which can be expressed as: ; In the formula, represents the comprehensive loss function, represents the data-driven loss function, represents the physical constraint loss function, represents the boundary condition loss function, , and They represent the weights of the data-driven loss function, physical constraint loss function, and boundary condition loss function respectively.

[0036] Furthermore, if Figure 2 As shown, the PINN model is trained, and the weight and bias of the PINN model are adaptively adjusted by minimizing the comprehensive loss function to obtain a method for a trained PINN model, including the following steps: S31: Select an optimizer to train the PINN model, randomly initialize the PINN model weights, and calculate the data-driven loss function, physical constraint loss function, and boundary condition loss function respectively; S32: Calculate a comprehensive loss function according to the data-driven loss function, the physical constraint loss function and the boundary condition loss function; S33: Use the automatic differentiation tool to calculate the gradient of the comprehensive loss function to the PINN model parameters, and use the optimizer to update the weights and biases of the PINN model. Repeat the operation until the comprehensive loss function converges or reaches the specified number of training times to obtain the optimal parameters of the PINN model.

[0037] like Figure 4 As shown in the figure, in actual operation, the L-BFGS optimizer can be used to train the PINN model and randomly initialize the PINN model weights (Xavier initialization or He initialization can be used). Considering that the comprehensive loss function in the PINN model includes data-driven loss function, physical constraint loss function and boundary condition loss function, it is necessary to sample points from different data sources, such as randomly sampling N data points from the data collected from the dam monitoring point to calculate the data-driven loss function; randomly generating M physical constraint points from the physical space to calculate the physical constraint loss function; sampling K points from the boundary conditions to calculate the boundary condition loss function, and calculating the comprehensive loss function based on the data-driven loss function, physical constraint loss function and boundary condition loss function. Then use the automatic differentiation tool to calculate the gradient of the comprehensive loss function to the neural network parameters, and use the optimizer to update the weights and biases of the PINN model. Repeat the above steps until the comprehensive loss function converges or reaches the specified number of training times to obtain the optimal parameters of the PINN model.

[0038] Specifically, when sampling data points, N points need to be randomly sampled from the data collected from the dam monitoring points to calculate the data-driven loss. If the number of data points is limited, uniform sampling or random sampling can be used. For areas where monitoring points are densely distributed, stratified sampling can be performed to make the samples more evenly distributed in space. When sampling physical constraint points, M points need to be randomly generated from the physical space. These points are used to calculate the residual loss of the physical control equation. The sampling method can use random uniform sampling, Latin hypercube sampling, and important area sampling. When sampling boundary condition points, for fixed boundary points, uniform sampling is required at the fixed boundary at the bottom of the dam to ensure that each key boundary point can be covered; for water pressure boundary points, points should be generated along the boundary normal in the water pressure action area upstream of the dam.

[0039] Among them, Figure 3 As shown, in step S50, the prediction result is compared with the monitoring index to identify the abnormal area, and the method for abnormal diagnosis of the abnormal area includes the following steps: S51: Based on historical monitoring data, the confidence interval method is used to determine the safety threshold of deformation distribution and the safety threshold of deformation rate, and the trained PINN model is used to predict the predicted value of dam spatial deformation. Based on the predicted value of dam spatial deformation, the predicted value of dam spatial deformation rate is calculated; S52: calculating a normalized deformation index according to a safety threshold of deformation distribution and a predicted value of dam spatial deformation, and calculating a normalized rate index according to a safety threshold of deformation rate and a predicted value of dam spatial deformation rate; S53: Determine the dam safety score by weight coefficient method based on the normalized deformation index and the normalized rate index; S54: Trigger an early warning based on the dam safety score and physical residual classification.

[0040] Specifically, the predicted value of the dam spatial deformation is obtained by the PINN model: , calculate the predicted value of dam spatial deformation The time derivative of , based on this definition, the normalized deformation index NDI and the normalized rate index NVI The relative abnormal degree of deformation and the abnormal degree of deformation rate are determined by these two parameters, and the final dam safety score is determined by the weight coefficient method.

[0041] Among them, the calculation formulas of normalized deformation index and normalized rate index are: ; ; In the formula, NDI represents the normalized deformation index, represents the predicted value of the dam spatial deformation at a certain monitoring point, represents the safety threshold of deformation distribution, NVI represents the normalized rate index, represents the predicted value of the spatial deformation rate of the dam at a certain monitoring point, Indicates the safety threshold of deformation rate.

[0042] Based on the normalized deformation index and normalized rate index, the dam safety score is determined by the weight coefficient method. The calculation formula of the dam safety score is: ; In the formula, S represents the dam safety score, , Represents the weight coefficient. Dam safety score S Indicates the overall safety status of the dam, ranging from [0,100], and the scoring classification is : Normal, safety status is good; : Mild abnormality, enhanced monitoring is recommended; : Serious abnormality, possible structural problems, and the emergency plan needs to be activated.

[0043] Physical residual It is calculated from the residual of the physical control equation in the PINN model, reflecting the degree of deviation between the model output and the physical law. The specific classification criteria are: , the residual is small, which conforms to the physical law; , the residuals are moderate, and there may be local anomalies; , the residual is large, there is a significant abnormality or risk, the above represents the upper limit of normal residual, Indicates the lower severe residual bound.

[0044] This implementation method defines the combination conditions of the safety score and the physical residual, and formulates the corresponding warning level as shown in Table 1. The macro risk level is determined by the safety score S, and then the physical residual is used to determine the macro risk level. Fine-tune warning levels to reflect local anomalies.

[0045] Table 1

[0046] Furthermore, the visualization module can be used to display the three-dimensional deformation field, residual heat map and safety score map. The three-dimensional deformation field presents the three-dimensional deformation field of the dam in an intuitive way, showing the deformation distribution and rate; the residual heat map generates a heat map of the physical constraint residual to intuitively display the abnormal area; the safety score map shows the overall score of the dam and the score distribution of the local area. The user interaction module can also be used to implement functions such as data input, result query and early warning setting. Specifically, the user can input or upload new safety monitoring data through the interface; the result query specifically allows the user to select a specific area or time period to view the deformation field and safety score; the early warning setting specifically allows the user to adjust the early warning threshold or score weight according to actual needs.

[0047] The present invention provides a mechanism data collaboratively driven dam safety dynamic intelligent diagnosis method, which constrains the model output through a physical constraint loss function so that it strictly satisfies the physical control equation, and constrains the model output through a boundary condition loss function so that it meets fixed conditions and water pressure conditions, thereby ensuring that the output of the PINN model has physical meaning and engineering applicability, so that the model has the characteristics of strong generalization ability, can adapt to different working conditions, regions and engineering conditions, and can make reliable predictions for unknown areas without relying on large-scale training data.

[0048] Corresponding to the above method embodiment, as Figure 5 As shown, the embodiment of the present invention also provides a dam safety status intelligent diagnosis system driven by mechanism data collaboration, including: A data acquisition module, which is used to collect environmental quantity data, dam deformation data and their corresponding monitoring point coordinates; A model building module is used to build a PINN model, wherein the PINN model includes an input layer, a hidden layer and an output layer, receives deformation data, monitoring point coordinates and environmental quantity data through the input layer, uses the hidden layer to capture nonlinear features of the input data, and outputs the predicted value of the dam deformation component in the output layer; A model training optimization module, which is used to construct a comprehensive loss function and train the PINN model, and adaptively adjust the weights and biases of the PINN model by minimizing the comprehensive loss function to obtain a trained PINN model, wherein the comprehensive loss function includes a data-driven loss function, a physical constraint loss function, and a boundary condition loss function; The deformation prediction module is used to predict the dam deformation field using the trained PINN model; The abnormal diagnosis and early warning module compares the prediction results with the monitoring indicators to identify the abnormal area, performs abnormal diagnosis on the abnormal area, and triggers an early warning based on the abnormal diagnosis results.

[0049] The present invention provides a dam safety state intelligent diagnosis system driven by mechanism data collaboration. The comprehensive loss function constructed by the system includes a data-driven loss function, a physical constraint loss function and a boundary condition loss function. When the PINN model is trained, the weight and bias of the model are adaptively adjusted by minimizing the comprehensive loss function to achieve collaborative optimization of the physical constraint loss and the data-driven loss. Through the synergistic effect of the mechanism data, it is ensured that the model prediction follows the physical laws and can better fit the actual monitoring data, and finally achieves higher-precision dam safety state prediction and intelligent diagnosis. The system has the significant advantages of high reliability, strong practicality and convenient technology promotion. It provides an important guarantee for the long-term safety of reservoir dam projects and provides advanced technical support for the digital and intelligent development of water conservancy projects.

[0050] For the convenience of description, the above system is described by dividing the functions into various system modules. Of course, when implementing the present disclosure, the functions of each system module can be implemented in the same or multiple software and / or hardware.

[0051] The system of the above embodiment is used to implement the intelligent diagnosis method of dam safety status driven by mechanism data collaboration in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0052] Corresponding to the above method embodiment, an embodiment of the present invention further provides a computer device, including: A memory for storing computer programs; A processor is used to implement the steps of the dam safety status intelligent diagnosis method driven by the mechanism data collaboration when executing a computer program.

[0053] In the embodiment of the present invention, the processor may be a central processing unit (CPU), an application specific integrated circuit, a digital signal processor, a field programmable gate array or other programmable logic devices.

[0054] The processor can call the program stored in the memory. Specifically, the processor can execute the operations in the embodiment of the dam safety status intelligent diagnosis method driven by the mechanism data collaboration.

[0055] The memory is used to store one or more programs, which may include program codes, and the program codes include computer operating instructions.

[0056] In addition, the memory may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device or other volatile solid-state storage device.

[0057] Corresponding to the above method embodiment, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the intelligent diagnosis method for dam safety status driven by the above-mentioned mechanism data collaboration are implemented.

[0058] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program codes.

[0059] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0060] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0062] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. A dam safety status intelligent diagnosis method driven by mechanism data collaboration, characterized by: The following steps are involved: S10: Collect environmental quantity data, dam deformation data and their corresponding monitoring point coordinates; S20: constructing a PINN model, wherein the PINN model includes an input layer, a hidden layer and an output layer, receiving deformation data, monitoring point coordinates and environmental quantity data through the input layer, using the hidden layer to capture nonlinear features of the input data, and outputting the predicted value of the dam deformation component in the output layer; S30: constructing a comprehensive loss function and training the PINN model, adaptively adjusting the weights and biases of the PINN model by minimizing the comprehensive loss function to obtain a trained PINN model, wherein the comprehensive loss function includes a data-driven loss function, a physical constraint loss function, and a boundary condition loss function; S40: Use the trained PINN model to predict the dam deformation field; S50: Compare the prediction result with the monitoring index to identify the abnormal area, perform abnormal diagnosis on the abnormal area, and trigger an early warning according to the abnormal diagnosis result.

2. According to claim 1, a mechanism data collaboratively driven dam safety status intelligent diagnosis method is characterized by: In S20, the hidden layer is a multi-layer fully connected network, and the activation function of the hidden layer is ReLU or tanh.

3. The dam safety status intelligent diagnosis method driven by mechanism data collaboration according to claim 1 or 2 is characterized by: In S30, the method for constructing a comprehensive loss function includes: The comprehensive loss function is the weighted sum of the data-driven loss function, the physical constraint loss function, and the boundary condition loss function, which is expressed as: ； In the formula, represents the comprehensive loss function, represents the data-driven loss function, represents the physical constraint loss function, represents the boundary condition loss function, , and They represent the weights of the data-driven loss function, the physical constraint loss function, and the boundary condition loss function, respectively, where: ； ； ； In the formula, Indicates the number of monitoring points involved in the calculation, Indicates The spatial coordinates and time points of the monitoring points, It means that the PINN model The predicted value of the three-dimensional deformation component of each monitoring point, Indicates Measured values ​​of three-dimensional deformation components at each monitoring point; Indicates the number of physical constraint points involved in the calculation, represents the divergence of the stress tensor, represents the gradient operator, represents the stress tensor, represents body force; represents the fixed boundary condition loss function, represents the water pressure boundary condition loss function.

4. The dam safety status intelligent diagnosis method driven by mechanism data collaboration according to claim 3 is characterized by: In S30, the PINN model is trained, and the method of adaptively adjusting the weight and bias of the PINN model by minimizing the comprehensive loss function includes: S31: Select an optimizer to train the PINN model, randomly initialize the PINN model weights, and calculate the data-driven loss function, physical constraint loss function, and boundary condition loss function respectively; S32: Calculate a comprehensive loss function according to the data-driven loss function, the physical constraint loss function and the boundary condition loss function; S33: Use the automatic differentiation tool to calculate the gradient of the comprehensive loss function to the PINN model parameters, and use the optimizer to update the weights and biases of the PINN model. Repeat the operation until the comprehensive loss function converges or reaches the specified number of training times to obtain the optimal parameters of the PINN model.

5. The method for intelligent diagnosis of dam safety status driven by mechanism data collaboration according to claim 4 is characterized by: In S50, the prediction result is compared with the monitoring index to identify an abnormal area, and the method for abnormal diagnosis of the abnormal area includes: S51: Based on the historical monitoring data, the safety threshold of the deformation distribution and the safety threshold of the deformation rate are determined by using the confidence interval method, and the predicted value of the dam spatial deformation is predicted by using the trained PINN model, and the predicted value of the dam spatial deformation rate is calculated based on the predicted value of the dam spatial deformation; S52: calculating a normalized deformation index according to a safety threshold of deformation distribution and a predicted value of dam spatial deformation, and calculating a normalized rate index according to a safety threshold of deformation rate and a predicted value of dam spatial deformation rate; S53: Determine the dam safety score by weight coefficient method based on the normalized deformation index and the normalized rate index; S54: Trigger an early warning based on the dam safety score and physical residual classification.

6. The method for intelligent diagnosis of dam safety status driven by mechanism data collaboration according to claim 5 is characterized by: In S52, the calculation formulas of the normalized deformation index and the normalized rate index are respectively: ； ； In the formula, NDI represents the normalized deformation index, represents the predicted value of the dam spatial deformation at a certain monitoring point, represents the safety threshold of deformation distribution, NVI represents the normalized rate index, represents the predicted value of the spatial deformation rate of the dam at a certain monitoring point, Indicates the safety threshold of deformation rate.

7. The method for intelligent diagnosis of dam safety status driven by mechanism data collaboration according to claim 5 is characterized by: In S53, the calculation formula for the dam safety score is: ； In the formula, S represents the dam safety score, and represents the weight coefficient, NDI represents the normalized deformation index, NVI represents the normalized rate index.

8. A dam safety status intelligent diagnosis system driven by mechanism data collaboration, characterized by: include: A data acquisition module, which is used to collect environmental quantity data, dam deformation data and their corresponding monitoring point coordinates; A model building module is used to build a PINN model, wherein the PINN model includes an input layer, a hidden layer and an output layer, receives deformation data, monitoring point coordinates and environmental quantity data through the input layer, uses the hidden layer to capture nonlinear features of the input data, and outputs the predicted value of the dam deformation component in the output layer; A model training optimization module, which is used to construct a comprehensive loss function and train the PINN model, and adaptively adjust the weights and biases of the PINN model by minimizing the comprehensive loss function to obtain a trained PINN model, wherein the comprehensive loss function includes a data-driven loss function, a physical constraint loss function, and a boundary condition loss function; The deformation prediction module is used to predict the dam deformation field using the trained PINN model; The abnormal diagnosis and early warning module compares the prediction results with the monitoring indicators to identify the abnormal area, performs abnormal diagnosis on the abnormal area, and triggers an early warning based on the abnormal diagnosis results.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the 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 program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images