Integrated method, system and equipment for rockfill dam deformation analysis based on multi-disciplinary model integration

Abstract

The present invention relates to the fields of computer and hydraulic structure, and discloses an integrated method for deformation analysis of rockfill dam based on multi-professional model integration, including the following steps: S1, using parameter inversion analysis method to establish rockfill material constitutive model, and also discloses a rockfill dam deformation analysis system based on multi-professional model integration, including: parameter inversion analysis module, used to establish rockfill material constitutive model, and discloses rockfill dam deformation analysis equipment based on multi-professional model integration, including: cluster management host, which is responsible for resource allocation and management of the entire equipment network. The containerized WebSocket technology greatly simplifies the real-time data exchange and communication between containers, and in a complex data processing process, the output of step S1 can immediately become the input of step S2, and the output of step S2 can immediately become the input of step S3. The advantage of "one communication, persistent connection" based on WebSocket technology reduces the server communication load.

Description

Technical Field

[0001] The present invention relates to the fields of computers and hydraulic structures, and in particular to an integrated method, system and equipment for deformation analysis of rockfill dams based on multi-professional model integration. Background Art

[0002] In the construction of rockfill dams, it is crucial to ensure the structural safety of the dam, in which deformation control plays a core role. With the rapid development of monitoring technology and data acquisition platforms, the amount of data generated by monitoring has increased significantly. When using data-driven methods to process and analyze data, the richness of multi-source monitoring data provides data guarantee for the health monitoring of rockfill dams. At present, the deformation analysis methods for rockfill dam projects can be divided into three categories.

[0003] The first category includes time series prediction, which usually uses modern mathematical methods such as grey time series models, machine learning and neural networks. Machine learning algorithms can automatically learn deformation patterns from a large amount of monitoring data, reducing the need for manual intervention. By training models to identify trends and anomalies in the data, machine learning can effectively predict the future deformation behavior of rockfill dams. At the same time, machine learning models can adapt to the changing monitoring data and have continuous learning capabilities. The second category involves spatial interpolation techniques. The most commonly used methods include inverse distance weighting and ordinary kriging. Spatial interpolation methods can estimate the deformation between monitoring points and generate a continuous deformation field, making the deformation distribution easier to visualize and understand. Interpolation methods can provide accurate deformation estimates in local areas. The third category uses spatiotemporal models to estimate the overall deformation of rockfill dams. Spatiotemporal models can integrate temporal and spatial variables to provide a more comprehensive deformation analysis, thereby better understanding the dynamic development process of rockfill dam deformation. By integrating time series and spatial analysis, the model can predict the deformation state of the rockfill dam at future time points and realize long-term monitoring of the rockfill dam.

[0004] However, there are often many problems with a single method, and the performance of machine learning models is highly dependent on the quality and quantity of training data. Monitoring data often have problems such as large deviations, incompleteness, or excessive noise, and data problems will directly affect the accuracy and reliability of the model. Spatial interpolation methods are usually based on the assumption of spatial autocorrelation, that is, points close to each other have similar properties. However, in practice, the deformation of rockfill dams may be affected by a variety of complex factors, and the spatial distribution of these factors may not satisfy this assumption. When the monitoring points are sparse, the uncertainty of the interpolation results increases, which may lead to misunderstandings of the deformation field. The spatial interpolation effect is limited by the observation scale and has scale sensitivity. Different scales may lead to completely different interpolation results, which is particularly evident in large-scale rockfill dam structures. The deformation of rockfill dams is a dynamic process, which is affected by many factors. The spatiotemporal model needs to be continuously adjusted and optimized to accurately capture these dynamic changes. At the same time, the effectiveness of the model is limited by the spatiotemporal scale. Different observation time and spatial range may require different model configurations and parameter adjustments.

[0005] The deformation calculation of rockfill dams requires processing a large amount of monitoring data, including but not limited to the displacement, stress, strain and other parameters of the dam body, foundation and surrounding geological bodies. As the acquisition frequency and accuracy of monitoring data continue to increase, the amount of data is growing exponentially. The processing and analysis of these data places extremely high demands on computing resources. The deformation calculation model of rockfill dams is usually nonlinear, and multiple physical and mechanical processes need to be considered. The coupling of these factors makes the calculation model very complex, and usually relies on large-scale finite element analysis or machine learning methods. The finite element calculation method calculates specific volumes and has high requirements on the computer's memory and processing power. At the same time, in order to ensure the accuracy of the rockfill dam deformation calculation, finer grids and more computing nodes are needed in the machine model, which further increases the complexity of the calculation and the required resources. Therefore, serious overload problems often occur during the calculation process.

[0006] Since the deformation of rockfill dams has obvious temporal and spatial correlations, and the monitoring data has significant variability in time and space, it is difficult to describe them using a unified deformation analysis model. At the same time, professional model calculations usually require the processing of a large number of data points and complex data relationships, which makes the model computationally demanding and the computational cost huge. Therefore, deformation assessment and prediction based on a single method that relies solely on a single-machine system still needs further improvement. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention provides an integrated method, system and equipment for rockfill dam deformation analysis based on multi-professional model integration, which solves the problems of insufficient data processing capability, low analysis accuracy and poor model adaptability existing in the single method in rockfill dam deformation analysis.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: An integrated method for rockfill dam deformation analysis based on multi-professional model integration includes the following steps:

[0009] S1. Establish the constitutive model of rockfill materials using parameter inversion analysis method;

[0010] S2. Perform finite element calculation based on the constitutive model to generate a deformation field of the rockfill dam;

[0011] S3. Reconstruct the deformation field of the rockfill dam using the Bayesian Kriging interpolation method.

[0012] Preferably, the steps of the parameter inversion analysis method specifically include the following steps:

[0013] Select agent computing models for training, including ANN neural network, support vector machine SVR, gradient boosting machine XGBOOST and random forest RF;

[0014] The non-dominated sorting genetic algorithm is used to perform parameter inversion and obtain the constitutive model of rockfill materials.

[0015] Preferably, the finite element calculation step specifically includes the following steps:

[0016] Obtain constitutive model parameters through containerized WebSocket full-duplex communication;

[0017] The containerized ABAQUS platform is used to perform finite element calculations of the deformation field of the rockfill dam.

[0018] Preferably, the steps of the Bayesian Kriging interpolation method specifically include the following steps:

[0019] Obtain finite element calculation results through containerized WebSocket full-duplex communication;

[0020] The deformation field of the rockfill dam is reconstructed by Bayesian Kriging interpolation.

[0021] The rockfill dam deformation analysis system based on multi-disciplinary model integration includes:

[0022] Parameter inversion analysis module, used to establish the constitutive model of rockfill materials;

[0023] Finite element calculation module, used to calculate the deformation field based on the constitutive model;

[0024] The deformation field reconstruction module is used to reconstruct the deformation field by Bayesian Kriging interpolation;

[0025] Data processing unit, used to process and cache data.

[0026] Preferably, the system is designed based on a microservice architecture and is deployed in a containerized manner. The parameter inversion analysis module uses a multi-objective strategy to perform parameter inversion, and the finite element calculation module is implemented by containerizing and packaging the ABAQUS platform and performing secondary development.

[0027] Preferably, the data processing unit comprises:

[0028] The processor is responsible for executing the core computing tasks of rockfill dam deformation analysis, including parameter inversion analysis of constitutive models, finite element calculations, and reconstruction of deformation fields, and establishing WebSocket full-duplex communication between the client and the algorithm module and database;

[0029] The cache is connected to the processor and is used to temporarily store calculation data and intermediate results, improve the quality of data transmission, ensure the stability of the calculation process, and reduce the risk of system overload.

[0030] Rockfill dam deformation analysis equipment based on multi-disciplinary model integration, including:

[0031] Cluster management host, which is responsible for resource allocation and management of the entire equipment network, and is also responsible for managing the methods and data allocation in the rockfill dam deformation analysis system based on multi-professional model integration, and supports adding existing algorithm models;

[0032] The data management host is connected to the cluster management host and is responsible for the persistent storage of computational data and historical data.

[0033] The algorithm management host is connected to the cluster management host and is responsible for managing and executing various algorithms involved in the system.

[0034] The readable storage medium of the rockfill dam deformation analysis equipment based on multi-professional model integration is applied to the data management host.

[0035] A chip system includes a processor for reading and executing a system program stored in a memory.

[0036] The present invention provides an integrated method, system and device for rockfill dam deformation analysis based on multi-professional model integration. It has the following beneficial effects:

[0037] 1. The present invention greatly simplifies the real-time data exchange and communication between containers through the containerized WebSocket technology. WebSocket allows containers to send and receive data in real time, which is conducive to the use of the rockfill dam deformation analysis model system with frequent interactions. In addition, in a complex data processing flow, the output of step S1 can immediately become the input of step S2, and the output of step S2 can immediately become the input of step S3. The advantage of "one-time communication, persistent connection" based on the WebSocket technology reduces the server communication load.

[0038] 2. The present invention enhances the flexibility of the platform and the convenience of user operation by supporting users to upload existing rockfill dam constitutive models locally or directly upload finite element calculation results, so that users can flexibly select data sources and calculation methods according to actual needs, optimize the calculation process, reduce the difficulty of non-professionals in performing finite element calculations of rockfill dams, and ensure the accuracy and reliability of the calculation results.

[0039] 3. The present invention can flexibly call the calculation results of each integrated module, automatically input the continuous calculation results and customize the input of the user's existing files, effectively integrate and utilize the existing computing resources and data, and improve the consistency and efficiency of the calculation.

[0040] 4. The present invention significantly improves the accuracy and efficiency of deformation analysis through a highly integrated calculation process, and significantly improves the efficiency and reliability of the analysis through a systematic processing process.

[0041] 5. The present invention can automatically allocate the required computing resources according to the complexity of the deformation analysis task through the advanced resource management strategy of the cluster management host, and proposes an adaptive strategy based on the Kafka message queue fusion VPA+HPA. For modules with large computing resource occupancy or access volume, the host will automatically schedule more computing resources, reducing delays and computing waste caused by improper resource allocation, and ensuring the stability of complex model calculations and finite element analysis.

[0042] 6. The present invention can efficiently process a large amount of calculation data from rockfill dam deformation analysis through the data management host, and through its efficient storage and fast access capabilities, it ensures the speed and accuracy of calculation data processing, making large-scale data analysis faster and more reliable.

[0043] 7. The present invention uses containerization technology through the algorithm management host to allow rapid deployment and updating of rockfill dam deformation analysis algorithms, such as finite element model and Kriging interpolation algorithm. These algorithms can be run in an independent and isolated environment, which increases the security and portability of the calculation process and also facilitates the testing and optimization of the algorithm.

[0044] 8. The present invention decouples algorithms and data on different hosts to enhance system flexibility and scalability, optimize cost-effectiveness, improve fault tolerance and reliability, and simplify data security and compliance measures. In addition, the decoupled architecture helps optimize performance and facilitate disaster recovery, while also making system management and maintenance more efficient and convenient. This architecture is particularly suitable for rockfill dam deformation analysis calculations that require processing large-scale data and complex computing tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flow chart of the method of the present invention;

[0046] Figure 2 It is the overall flow chart of the method of the present invention;

[0047] Figure 3 It is a schematic diagram of the deformation calculation process of the rockfill dam deformation analysis system of the present invention;

[0048] Figure 4 This is a schematic diagram of the WebSocket service containerization process of the present invention;

[0049] Figure 5 A schematic diagram of the execution flow of the parameter inversion analysis method encapsulated by the system of the present invention;

[0050] Figure 6 A schematic diagram of the execution flow of the finite element calculation method for the system package of the present invention;

[0051] Figure 7 A schematic diagram of the execution flow of the rockfill dam deformation field reconstruction method encapsulated by the system of the present invention;

[0052] Figure 8 is a schematic diagram of the device of the present invention;

[0053] Fig. 9 A diagram of the cluster management host management architecture of the device of the present invention;

[0054] Fig.10 This is a diagram of the cluster management host load balancing architecture of the present invention;

[0055] Fig.11 It is a schematic diagram of the load balancing result of the present invention;

[0056] Fig.12 This is a data management architecture diagram of a data management host of the present invention;

[0057] Fig.13 It is a flow chart of algorithm deployment of the algorithm management host of the present invention;

[0058] Fig.14 This is a diagram of the algorithm management architecture of the algorithm management host of the present invention;

[0059] Fig.15 It is a framework diagram of the system of the present invention;

[0060] Fig.16 A flowchart for the system and device execution of the method of the present invention;

[0061] Fig.17 This is a flow chart of Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] Please see attached Figure 1 -Attached Figure 4 The embodiment of the present invention provides an integrated method for rockfill dam deformation analysis based on multi-professional model integration, comprising the following steps:

[0064] S1. Establish the constitutive model of rockfill materials using parameter inversion analysis method;

[0065] S2. Perform finite element calculation based on the constitutive model to generate the deformation field of the rockfill dam;

[0066] S3. Reconstruct the deformation field of the rockfill dam using the Bayesian Kriging interpolation method.

[0067] Specifically, the present invention efficiently integrates the above three rockfill dam deformation analysis processes through containerized integration technology, realizes mutual communication and result fusion of various algorithm processes and tool components, aims to solve the limitations of the existing technology in the single means of rockfill dam deformation analysis, adopts full-duplex communication combined with containerized technology and configures load balancing strategy to achieve efficient integration and high-performance management of professional models related to various stages of rockfill dam deformation analysis.

[0068] The present invention combines WebSocket full-duplex communication technology with containerization technology and container orchestration tools to manage the integration and operation of various system models, simplifying the integration difficulty of various algorithm models and establishing data communication between various models. The present invention deploys the WebSocket service in one or more containers. It ensures that the WebSocket service can run in an isolated environment to ensure stability and security. The container is configured to automatically run the WebSocket service code at startup.

[0069] The present invention first performs inverse calculation on the physical properties of the rockfill material through the S1 step model to obtain its accurate constitutive model. This step provides the physical accuracy and calculation reliability of the model for subsequent finite element calculation.

[0070] Afterwards, the constitutive model obtained by inversion is used to simulate the overall deformation structure of the rockfill dam through step S2 of the present invention. This not only helps to predict the response of the rockfill dam under various working conditions, but also can guide decision-making during engineering design and construction.

[0071] Finally, the deformation field of the rockfill dam is reconstructed using step S3 of the present invention, allowing for a more accurate analysis and prediction of the deformation behavior of the rockfill dam.

[0072] WebSocket full-duplex communication is used in the communication process of each module. Figure 3 The figure shows the process of establishing containerized WebSocket full-duplex communication.

[0073] In general, the overall process of the present invention is as follows:

[0074] In the S1 method calculation, the S1 calculation task related parameters are transferred to the Kafka cache queue through WebSocket full-duplex communication. The multiple Kafka consumer nodes of S2 on the algorithm management host obtain the calculation task parameters and calculate the rockfill dam constitutive model. The cluster management host increases or decreases the number of S1 computing nodes according to the actual computing needs, and stores the obtained constitutive model persistently to the data management host.

[0075] In the calculation of the S2 method, the S1 calculation constitutive model file is obtained from the data management host through WebSocket full-duplex communication and imported into the method; in the calculation of the S2 method, the S2 calculation task related parameters are passed into the Kafka cache queue through WebSocket, and the multi-Kafka consumer nodes of S2 on the algorithm management host obtain the calculation task parameters and calculate the rockfill dam deformation field file. The cluster management host increases or decreases the number of S2 calculation nodes according to the actual calculation needs, and stores the obtained rockfill dam deformation field file persistently to the data management host;

[0076] In the S3 method calculation, the S2 calculation file of the rockfill dam deformation field is obtained from the data management host through WebSocket full-duplex communication and imported into the calculation method; in the S3 method calculation, the S3 calculation task related parameters are passed into the Kafka cache queue through WebSocket, and the S3 multi-Kafka consumer nodes on the algorithm management host obtain the calculation task parameters and calculate the rockfill dam deformation field reconstruction file. The cluster management host increases or decreases the number of S3 computing nodes according to the actual computing needs, and the obtained rockfill dam deformation field reconstruction file is persistently stored in the data management host.

[0077] The calculation S1 of the constitutive model of rockfill dam is the first step in the deformation analysis of rockfill dam. In the present invention, the first step of calculating the constitutive model of rockfill dam needs to provide the inversion parameters of the constitutive model of rockfill dam. The parameters that need to be imported in the present invention mainly select the parameters that have a greater influence on the stress and deformation of the dam body and have a greater uncertainty obtained through conventional triaxial tests. The selected parameters include the tangent modulus K, the tangent modulus index n, and the bulk modulus K in the EB model. b , bulk modulus index m; α rheological rate in Merchant’s seven-parameter model, b, c, d rheological deformation parameters, and c in the three-parameter wetting model w 、n w , b w The model parameters are taken as the parameters to be inverted.

[0078] Please see attached Figure 3 and attached Figure 5 , the steps of the parameter inversion analysis method specifically include the following steps:

[0079] Select agent computing models for training, including ANN neural network, support vector machine SVR, gradient boosting machine XGBOOST and random forest RF;

[0080] The non-dominated sorting genetic algorithm is used to perform parameter inversion and obtain the constitutive model of rockfill materials.

[0081] Specifically, the back calculation of rockfill dam parameters consists of three steps:

[0082] Step 1: The system implements the inversion calculation of rockfill dam parameters. The calculation process includes the accuracy test of the proxy model and the inversion calculation of parameters. The user selects the proxy calculation model and inputs the calculation file, which includes the model training file and the model test file. After clicking the calculation, the system background starts the asynchronous calculation of the proxy model. The asynchronous calculation accuracy results can be transmitted to the front-end chart in real time through the WebSocket full-duplex communication built by the system to realize the visualization of the model accuracy. Among them, the optimal proxy model of each partition needs to be calculated for the overall calculation of the rockfill dam. In order to provide system robustness, the system automatically detects the calculated dam area model to avoid repeated model calculations in the same dam area. Each proxy model must meet the calculation results of all dam sections before the inversion calculation can be performed.

[0083] Step 2 is to input the parameter inversion calculation model. First, input the user's basic information to facilitate result retrieval. According to the calculation proxy model calculated in step 1, input the proxy model accuracy, and use the non-dominated sorting genetic algorithm NSGA-Ⅲ, and input the measurement point deformation file. The system provides a user-defined interactive interface for genetic algorithm parameters, including the number of particle swarms and the number of iterations, where the number of iterations is positively correlated with the system model calculation time.

[0084] Step 3, the robustness calculation and verification process of the system. Before the calculation starts, you need to click to check the inversion calculation conditions. The system will feedback the missing data to the user. After the conditions are met, the calculation function can be opened and run normally. Click the operation system to display the calculation process. After the calculation is completed, the system outputs the parameter inversion calculation model parameters and marks the completion of the inversion calculation. At the same time, the system will store the obtained rockfill dam constitutive file permanently to the data management host.

[0085] Finite element calculation S2 of rockfill dam deformation field is the second step of rockfill dam deformation analysis. The system finite element calculation management module is packaged and secondary developed based on ABAQUS, such as Figure 6 As shown, users can publish finite element calculation tasks in the system. Compared with directly using the complex operation interface of the finite element calculation software, the system can simplify the finite element calculation to the greatest extent and lower the threshold for using the finite element software.

[0086] Please see attached Figure 3 and attached Figure 6 , the steps of finite element calculation specifically include the following steps:

[0087] Obtain constitutive model parameters through containerized WebSocket full-duplex communication;

[0088] The containerized ABAQUS platform is used to perform finite element calculations of the deformation field of the rockfill dam.

[0089] Specifically, the finite element calculation task of rockfill dam is submitted in four steps:

[0090] Step 1: When the finite element calculation management module starts calculation, it automatically establishes WebSocket full-duplex communication with the background to realize the automatic input of the calculation result file of the first step algorithm as the calculation parameter of the second step. Users can also choose to upload the existing rockfill dam constitutive model to the finite element calculation management module for finite element calculation, enter the monitoring data deadline and click Submit. After the system prompts that the task has been submitted, step 2 can be executed.

[0091] Step 2: Input of text information of finite element calculation. The user submits the calculation task number according to step 1 and inputs the text explanation of the parameter information related to the finite element calculation.

[0092] Step 3: The calculation task starts. On the premise that steps 1 and 2 are submitted normally, the user enters the custom task ID according to the task number, and clicks "Calculate" to start the finite element calculation. The calculation task running status is divided into 4 categories, where green means the task is completed, yellow means the task is in progress, blue means the task is idle, and red means the system calculation delay. After the calculation starts, the button status changes from blue to yellow. After the calculation is completed, the normal calculation is green, and red means calculation delay (that is, error calculation).

[0093] Step 4 is the extraction of finite element calculation results. Finite element calculation can complete the complete step extraction. According to the task number, the user enters the calculation extraction step according to the monitoring data time and constitutive file name of steps 1 to 3, and clicks Submit. The task can be executed in the background. Users can download the result file of the finite element calculation step extraction. It also supports users to upload existing finite element result files.

[0094] This module can visualize the uploaded data in three-dimensional charts. The three-dimensional charts use color grading to show the deformation degree of the rockfill dam calculated or uploaded results.

[0095] The rockfill dam deformation field reconstruction calculation S3 is the third step of the rockfill dam deformation analysis. It adopts a rockfill dam deformation field reconstruction method based on Bayesian Kriging interpolation. This step is applied to the rockfill dam deformation analysis system. The method adopts a loosely coupled integration strategy to modularize the algorithm input, processing and output to ensure the algorithm's ease of use and intuitive operation. This step uses the Bayesian Kriging interpolation rockfill dam deformation field reconstruction method to calculate the reconstructed deformation field of the rockfill dam as the final result of the rockfill dam deformation analysis calculation.

[0096] Please see attached Figure 3 and attached Figure 7 , the steps of the Bayesian Kriging interpolation method specifically include the following steps:

[0097] Obtain finite element calculation results through containerized WebSocket full-duplex communication;

[0098] The deformation field of the rockfill dam is reconstructed by Bayesian Kriging interpolation.

[0099] Specifically, the three steps of calculation of the system integrated deformation field reconstruction module are as follows:

[0100] Step 1: Input calculation parameters including calculation period and monitoring time, and input finite element calculation file and monitoring settlement value file to be reconstructed. When the deformation field reconstruction module starts calculating, it automatically establishes WebSocket full-duplex communication with the background to realize automatic input of calculation result file in the second step, which is used as calculation file parameter in the third step. Finite element calculation file can also be input by the user.

[0101] Step 2: Select the deformation field reconstruction model integrated by the system. The system provides three types of models including Gaussian model, spherical model, and exponential model, and calculate the accuracy of each model.

[0102] Step 3: Enter the calculation period and monitoring time parameters to start the calculation. The system will provide real-time feedback on the calculation status to the user. Under the premise of multi-user parallel operation, the information box will include the name of the submitter and the model selection information. After the calculation is completed, the results are automatically transmitted to the client through the established full-duplex communication to display in the three-dimensional chart, and the model calculation accuracy is output. At the same time, the system will store the obtained rockfill dam deformation field reconstruction file permanently to the data management host.

[0103] Please see attached Fig.15 , a rockfill dam deformation analysis system based on multi-disciplinary model integration, including:

[0104] Parameter inversion analysis module, used to establish the constitutive model of rockfill materials;

[0105] Finite element calculation module, used to calculate the deformation field based on the constitutive model;

[0106] The deformation field reconstruction module is used to reconstruct the deformation field by Bayesian Kriging interpolation;

[0107] Data processing unit, used to process and cache data.

[0108] The system is designed based on a microservice architecture and uses containerized deployment. The parameter inversion analysis module uses a multi-objective strategy to perform parameter inversion, and the finite element calculation module is implemented through containerized packaging and secondary development of the ABAQUS platform.

[0109] Specifically, the system of the present invention integrates multiple professional analysis models based on the microservice architecture, and combines and applies multiple methods such as machine learning algorithms, spatial interpolation methods, and spatiotemporal models for different stages of rockfill dam deformation analysis. In this way, the present invention can combine the advantages of various models to improve the comprehensiveness and accuracy of rockfill dam deformation monitoring and analysis.

[0110] The data processing unit includes:

[0111] A processor, which is responsible for performing the core computing tasks of rockfill dam deformation analysis, including parameter inversion analysis of constitutive models, finite element calculations, and reconstruction of deformation fields;

[0112] The cache is connected to the processor and is used to temporarily store calculation data and intermediate results, improve the quality of data transmission, ensure the stability of the calculation process, and reduce the risk of system overload.

[0113] Specifically, the processor implements the communication between the integrated method computing and data management host and the algorithm management host through WebSocket full-duplex communication, realizing "one communication, persistent connection" and reducing the internal communication cost of the system.

[0114] The cache uses Kafka message queues to temporarily store the data submitted by the calculation tasks of the integrated method for rockfill dam deformation analysis based on multi-professional model integration, reduce the high load of the system under peak value calculation, improve data access speed, support fast calculation processing, and reduce the response time of the system. Among them, the message queue, as the middleware between the client and the server, can temporarily store the incoming requests, and then pass them to the back-end cluster for processing one by one according to the processing capacity, so as to smooth the load peak and avoid the cluster service from being overloaded due to the sudden increase in requests.

[0115] Please see attached Figure 8 -Attached Fig.14 , a rockfill dam deformation analysis device based on multi-professional model integration, including:

[0116] Cluster management host, which is responsible for resource allocation and management of the entire equipment network, and is also responsible for managing the methods and data allocation in the rockfill dam deformation analysis system based on multi-professional model integration, and supports adding existing algorithm models;

[0117] The data management host is connected to the cluster management host and is responsible for the persistent storage of computational data and historical data.

[0118] The algorithm management host is connected to the cluster management host and is responsible for managing and executing various algorithms involved in the system.

[0119] Specifically, the device is used to optimize and accelerate the model calculation of rockfill dam deformation analysis, realize containerized management of various professional models, and combine the load balancing method of message queue to ensure the calculation stability of each integrated model. Therefore, the device not only supports efficient data processing and complex algorithm calculations, but also provides a stable and reliable computing platform that can support multi-user use.

[0120] It mainly uses a container orchestration tool to achieve container management in each host. The cluster management host mainly includes the control plane, which mainly includes cluster management components, cluster scheduling components, cluster control components and ETCD database. By integrating efficient scheduling algorithms and load balancing mechanisms in the cluster management host, resource management and load balancing of the data management host and algorithm management host in the cluster are achieved. The data management host and algorithm management host mainly include proxy components, network proxy components, and container engines.

[0121] When a user submits a new algorithm or data, the cluster management host will deploy the Pod to the algorithm management host, and the relevant data will be deployed to the data management host. Deploying different types of workloads on different hosts can achieve resource isolation, which helps prevent an overloaded workload from affecting another workload. Intensive data processing tasks will not interfere with algorithm calculations.

[0122] The cluster management host collects resource usage data of all nodes and Pods through MetricsServer, such as resource access and memory usage. The data is collected through the / metrics / resource interface of Kubelet. Pod is a collection of containers for K8s cluster content. In order to ensure the load balance of cluster resources, each Pod contains only one container. The dynamic scheduler uses preset rules and real-time monitoring data. Containers that generate burst traffic will be scheduled with more resources, and more Pod copies will be generated based on the access volume to optimize resource usage for each module. The number of Pods is automatically adjusted according to the real-time load and preset performance indicators to ensure load balance on each node and improve the overall performance and stability of the cluster.

[0123] The cluster management host first receives the Pod configuration information submitted by the user through the API Server, and then stores the cluster status through ETCD. Fig. 9 and attached Fig.10 As shown in the figure, the auto-scaling controller monitors the resource usage of the Pod, executes the adaptive strategy based on the Kafka message queue fusion VPA+HPA, and dynamically adjusts the Pod computing resources and the number of Pod replicas according to the set thresholds. The Kafka consumer and model algorithm have been deployed in the Pod to ensure processing efficiency and response speed.

[0124] Implementation steps:

[0125] Step 1: In the present invention, Kafka message queue is selected to be deployed outside the K8s cluster, and persistent storage and high availability configuration are set inside the K8s cluster. At the same time, a queue is created to manage computing task messages, and message retention policies and consumer groups are set. The Kafka message queue is used as an intermediate cache from the client to the server;

[0126] Step 2: Deploy the Kafka consumer module and model algorithm in the Pod node of the algorithm management host of the K8s cluster at the same time, import the computing task data sent by the Kafka consumer module receiving queue into the model algorithm, pull the computing task messages from the message queue, and manage the computing task resources through VPA+HPA;

[0127] Step 3: Specify the Dependency of the monitoring computing task service and set updateMode to Auto in the VPA configuration file;

[0128] Step 4: Use VPA+HPA to automatically adjust the resource request and quantity of Pods based on the load.

[0129] The data management host is responsible for managing all data in the cluster and has high-speed data processing capabilities to meet high-concurrency data operations. In order to ensure that the data management host has high-speed data processing capabilities and fast data access, the present invention is equipped with a server with a powerful CPU and a large amount of RAM and a large storage SSD to provide faster read and write speeds. At the same time, the network interface card supports 50GbE or above to meet the needs of high-concurrency data transmission. Fig.12 As shown in the figure, the data management host establishes NFS as a data storage node, allowing multiple clients to access the same file system through the network to implement shared data applications. In a cluster environment, multiple Pods can read and write the same file system at the same time without worrying about data synchronization and data consistency issues.

[0130] The readable storage medium of the rockfill dam deformation analysis device based on multi-professional model integration is applied to the rockfill dam deformation analysis device based on multi-professional model integration.

[0131] Each step of the rockfill dam deformation analysis requires batch processing of input data, which is then analyzed using an integrated algorithm through the system module and the results are stored in an NFS shared volume, making it convenient for subsequent steps to analyze and visualize the results. The implementation steps are as follows:

[0132] Step 1: Create a PersistentVolume in the cluster and point it to the NFS server. Build a container image for the algorithm model.

[0133] Step 2: Create a PersistentVolumeClaim (PVC) and ensure that the algorithm can use the PVC.

[0134] Step 3: Deploy a Pod that contains a data analysis tool, which includes a customized algorithm and commonly used data analysis software packages.

[0135] Step 4: Run data analysis, load data from external sources or internal databases, perform analysis through integrated algorithms, and finally output the processed results to the mounted NFS storage / data.

[0136] All calculation outputs are centrally managed and stored on the data management host, which not only improves the security and accessibility of data, but also makes data backup and recovery simple and efficient. This architecture guarantees the needs of rockfill dam deformation analysis for processing large amounts of data and generating large amounts of output.

[0137] The algorithm management host has efficient computing power to process algorithm operations, support rapid deployment and update of algorithms, and achieve isolation of algorithm environments. Fig.13As shown in the figure, the host generates an image for each algorithm and its necessary environment, generates a container based on the image and deploys it to the production environment. At the same time, the image is uploaded to the remote image warehouse for storage. Each Pod contains only one container, which contains professional algorithms and Kafka consumer function modules. In order to support efficient machine learning algorithm calculations, the host is equipped with a professional GPU to support algorithm training. The host only runs algorithm-related Pods. Fig.14 As shown in the figure, Dep l oyment is used to deploy and manage algorithm applications. Each algorithm is allocated 20 Pods. In the subsequent computing task submission, the number of Pods is automatically increased or decreased based on the resource usage and algorithm access requirements of different algorithms, based on the resource allocation mechanism of the cluster management host. At the same time, the algorithm management host is equipped with an algorithm update and rollback mechanism to ensure algorithm updates and fault tolerance.

[0138] The deformation analysis of rockfill dams requires the use of deep learning algorithms such as image recognition to quickly identify cracks on the dam surface. This algorithm needs to process highly concurrent image uploads and quickly give classification results.

[0139] The implementation steps are as follows:

[0140] Step 1: Deploy the machine learning model and create a container image that contains the machine learning model. The image contains all the necessary code and libraries, such as TensorFlow or PyTorch.

[0141] Step 2: Create a K8sDep l oyment at the same time and deploy the model in the algorithm management host.

[0142] Step 3: Create Service and Ingress to ensure external access to the service.

[0143] Step 4: Send the image to the / recogn i ze endpoint of the service through the Websocket protocol. The service will run the rockfill dam body image recognition model on the GPU-accelerated node and quickly return the classification result.

[0144] The readable storage medium of the rockfill dam deformation analysis equipment based on multi-professional model integration is applied to the data management host.

[0145] A chip system includes a processor for reading and executing a system program stored in a memory.

[0146] Embodiment 1:

[0147] Please see attached Fig.16 and attached Fig.17 In an optional example, the present invention is applied in combination with a deformation analysis project of a high core rockfill dam.

[0148] Step 1: Please refer to the attached Figure 5 , Attachment Fig.16 and attached Fig.17 , the constitutive parameters of the rockfill dam are calculated through the integrated method S1 process of the present invention.

[0149] The dam filling material contractor and the triaxial test results are input, and the value ranges of the parameters to be inverted in each zone are determined as shown in the table.

[0150]

[0151]

[0152] Considering that there are many target measurement points in each partition and the training sample size is large, the support vector machine model has low processing efficiency for large-scale training sets. Therefore, in this example, three machine learning algorithms, artificial neural network, gradient boosting machine and random forest, are used to train the proxy model in step 1 of S1. The system can automatically input the verification evaluation indicators, among which the smaller the mean absolute percentage error MAPE, the root mean square error RMSE or the correlation coefficient R 2 The larger the value, the higher the accuracy of the training model and the stronger the generalization ability. After step 1, the calculation results are as follows:

[0153]

[0154] In the second step of the S1 process, the NSGA-Ⅲ algorithm is selected for iterative optimization. The input parameters in the system are as follows. In this example, the initial population size of the algorithm is set to 500, the maximum evolutionary generation is set to 1000, the mutation probability is set to 0.1, and the crossover and mutation parameters are set to 30 and 20 respectively. The multi-objective optimization parameters finally obtained and the optimal constitutive model of the rockfill dam are calculated as shown in the following table.

[0155]

[0156]

[0157] Step 2: Please refer to the attached Figure 6 , Attachment Fig.16 and attached Fig.17 , the finite element deformation field of the rockfill dam is calculated through the integrated method S2 process of the present invention.

[0158] The calculation result of step 1 is used as the calculation data of step 2, and is directly imported from the system. After inputting the basic text information of finite element calculation (which can be ignored), click calculation according to the task number. The calculation button changes from blue to yellow to indicate the start of calculation, and the button changes to green to indicate the completion of finite element calculation. The finite element calculation of a high core wall rockfill dam is divided into 146692 grids. The calculation results are shown in the table:

[0159]

[0160]

[0161] Step 3: Please refer to the attached Figure 7 , Attachment Fig.16 and attached Fig.17 , the finite element deformation field reconstruction of the rockfill dam is calculated through the integrated method S3 process of the present invention.

[0162] It is difficult to fully consider the various engineering characteristics of rockfill materials during deformation through the macroscopic constitutive model of the rockfill body, and it is also difficult to accurately determine all model parameters. Therefore, the deformation field of the rockfill dam calculated by the finite element module based on the constitutive model calculated in step 2 of the integrated method needs to be further optimized and reconstructed.

[0163] The calculation results of step two are used as the calculation data of step three, and are directly imported from the system. At the same time, the finite element calculation file and the monitoring settlement value file that need to be reconstructed are input, and the deformation field reconstruction model integrated by the system is selected. The system provides three types of models including Gaussian model, spherical model, and exponential model. The deformation field reconstruction model is calculated three times. The calculation results of the three models are compared, and the optimal model is selected according to the returned accuracy.

[0164] The calculation accuracy of each model is shown in the following table:

[0165] Selecting a computational model Model fit mean residuals Mean Kriging Variance Gaussian Model 10793.20 98309.60 Spherical Model 11643.25 106543.75 Exponential Model 13030.64 121536.31

[0166] Finally, the Gaussian model was selected as the deformation field reconstruction calculation model, and the calculation period and monitoring time parameters were input to start the calculation. The deformation field reconstruction results are shown in the table:

[0167]

[0168]

[0169] Experimental summary:

[0170] The final result uses the four-pronged verification method, randomly selects 25% of the monitoring points as verification points, and uses the average error and root mean square error indicators to evaluate the comprehensive error of the measuring points. The error comparison is shown in the table:

[0171] method MAE(cm) RMSE Integration method of the present invention 7.436 14.231 Single inversion parameter calculation method 39.742 46.713 Original triaxial test parameter calculation method 132.341 144.394

[0172] The results show that the accuracy of the reconstructed deformation field of the rockfill dam calculated by the integrated method of rockfill dam deformation analysis based on multi-professional model integration has been greatly improved compared with a single method, and the calculation results more accurately reflect the actual deformation of the rockfill dam.

[0173] Embodiment 2:

[0174] When the computing task of the rockfill dam computing system is submitted during the emergency period, two users are simulated to submit multiple computing tasks at the same time. The background uses the Kafak message queue and VPA+HPA to coordinate and handle high-traffic events to solve the system overload problem caused by the submission of a large number of tasks during the emergency period. Please refer to the attached Fig.10 , the load balancing results of the algorithm management host are as follows Fig.11 shown.

[0175] The results of the balanced strategy execution are as follows:

[0176]

[0177]

[0178] Experimental summary:

[0179] From the above, it can be concluded that the load balancing strategy under load synergy proposed by the present invention can effectively handle the high load problem in each computing node when the amount of computing tasks is large. By decoupling the user operation and computing task processing process, and the message queue buffering task data under high load, combined with VPA automatically adjusting the Pod resources of the algorithm model node and HPA automatically adjusting the number of Pod copies of the algorithm model node. It is guaranteed that when the demand for algorithm model computing resources increases, more CPU and memory resources and more Pod copy data can be obtained to handle multi-user high resource load computing tasks.

[0180] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An integrated method for rockfill dam deformation analysis based on multi-disciplinary model integration is characterized by: The following steps are involved: S1. The non-dominated sorting genetic NSGA-III algorithm is used for parameter inversion optimization. The model parameters are optimized through an evolutionary search mechanism. At the same time, a variety of machine learning agent models are integrated to form an efficient iterative feedback mechanism to improve the convergence speed and global search capability of parameter optimization, so that the agent model can be adaptively optimized during the inversion process of the constitutive model. The NSGA-III algorithm is deployed through containerized packaging to realize batch calculation of each generation of population on multiple computing nodes and perform fitness evaluation; S2. Based on the constitutive model, the containerized ABAQUS finite element calculation platform is used to realize efficient calculation of the rockfill dam deformation field, achieve high portability and scalability of the computing environment, and realize real-time interaction and efficient data transmission of the parameter inversion process through containerized WebSocket full-duplex communication, forming a deep integration architecture of the computing module and the communication module, and improving the automation degree and computing efficiency of deformation calculation; S3. By adapting the Bayesian Kriging interpolation method through containerized encapsulation, all deformation field reconstruction calculation modules are independently deployed in the container environment. Based on the containerized message queue and load balancing mechanism, the automatic dynamic call of the interpolation calculation process is realized to ensure the high resource demand of the model in the calculation task, thereby improving the efficiency of deformation reconstruction, calculation stability and scalability, and realizing fully automated rockfill dam deformation reconstruction calculation.

2. The integrated method for rockfill dam deformation analysis based on multi-professional model integration according to claim 1 is characterized in that: The steps of the parameter inversion analysis method in step S1 specifically include the following steps: A machine learning agent model was constructed based on artificial neural network, support vector regression, gradient boosting machine and random forest, and inversion parameter optimization search was performed by integrating non-dominated sorting genetic NSGA-Ⅲ algorithm; The Pareto optimal solution is selected through non-dominated sorting and crowding calculation, and genetic operations are performed; The two are integrated to automatically update the proxy model during the optimization process, improve the inversion accuracy and calculation efficiency, and obtain a high-precision constitutive model of the rockfill dam; The NSGA-III algorithm module is encapsulated through containerization technology for multi-parallel computing, and each generation of population is calculated in batches to achieve fitness evaluation of multiple computing instances at the same time.

3. The integrated method for rockfill dam deformation analysis based on multi-professional model integration according to claim 1 is characterized in that: The finite element calculation step in step S2 specifically includes the following steps: Based on containerized WebSocket full-duplex communication, dynamic interaction and real-time transmission of constitutive model parameters are realized to ensure efficient collaboration between the computing module and the communication module; Using the containerized ABAQUS finite element computing platform, the deformation field calculation of the rockfill dam is performed in an independently packaged computing environment, achieving efficient parallel computing and cross-platform portability; Through the seamless connection between WebSocket components and ABAQUS containers, a deeply integrated architecture of computing and communication is built to achieve automated parameter transmission, computational scheduling, and result feedback, thus improving the intelligence and real-time performance of deformation calculations.

4. The integrated method for rockfill dam deformation analysis based on multi-professional model integration according to claim 1 is characterized in that: The step of adapting the Bayesian Kriging interpolation method by container encapsulation in step S3 specifically includes the following steps: Based on containerized WebSocket full-duplex communication, real-time acquisition and efficient transmission of deformation calculation results are achieved, ensuring low latency and stability of data interaction; The Bayesian Kriging interpolation method is fully deployed in an independent container environment to achieve standardized encapsulation of computing modules and support remote calls and automated computing. Use the message queue mechanism to manage interpolation calculation tasks to ensure the automated execution, real-time response, and dynamic optimization of the calculation process; The load balancing mechanism supports high-resource scheduling of interpolation models, ensuring sufficient allocation of computing resources for interpolation calculations and improving the reconstruction accuracy and computational stability of the deformation field.

5. The rockfill dam deformation analysis system based on multi-professional model integration is characterized by: include: Parameter inversion analysis module, which is used to integrate the non-dominated sorting genetic NSGA-Ⅲ algorithm and the machine learning agent model to establish the rockfill material constitutive model and perform efficient parameter optimization; The finite element calculation module is based on the containerized ABAQUS platform and uses efficient parallel computing technology to achieve finite element solution of the rockfill dam deformation field; The deformation field reconstruction module supports the high computing resource scheduling requirements of the reconstruction module, and uses containerization technology combined with a load balancing mechanism to ensure high accuracy and stability of interpolation calculations; The data processing unit is used to efficiently process, cache and manage computing data, and supports real-time interaction and remote calls.

6. The rockfill dam deformation analysis system based on multi-professional model integration according to claim 5 is characterized in that: The system is designed based on a microservice architecture and adopts containerized deployment to improve the scalability and cross-platform adaptability of the system. Through deep integration of computing and communication modules and containerized WebSocket full-duplex communication, real-time interaction of parameter inversion, finite element calculation, and deformation field reconstruction is achieved, thereby improving data transmission efficiency and the overall computing performance of the system.

7. The rockfill dam deformation analysis system based on multi-professional model integration according to claim 5 is characterized in that: The data processing unit comprises: The processor is responsible for executing the core computing tasks of rockfill dam deformation analysis, including parameter inversion analysis of constitutive models, finite element calculations, and reconstruction of deformation fields, and establishing WebSocket full-duplex communication between the client and the algorithm module and database; The cache is connected to the processor and is used to temporarily store calculation data and intermediate results, improve the quality of data transmission, ensure the stability of the calculation process, and reduce the risk of system overload.

8. Rockfill dam deformation analysis equipment based on multi-professional model integration, characterized by: Used in conjunction with any one of the rockfill dam deformation analysis systems based on multi-professional model integration as described in claims 5 to 7, comprising: The cluster management host is responsible for resource allocation and load balancing management of the entire device network, and is also responsible for managing the method and data allocation in the system, and supports adding existing algorithm models; The data management host is connected to the cluster management host and is responsible for the persistent storage of computational data and historical data. The algorithm management host is connected to the cluster management host and is responsible for managing and executing various algorithms involved in the system.

9. A readable storage medium for rockfill dam deformation analysis equipment based on multi-professional model integration, characterized in that: Applicable to the data management host described in claim 8 above.

10. A chip system, characterized in that: Used in conjunction with the rockfill dam deformation analysis system based on multi-professional model integration as described in any one of claims 5 to 7 above, the system comprises a processor for reading and executing a system program stored in a memory.

Images