A high-rise building settlement prediction method, system, storage medium and program product

Abstract

A method, system, storage medium and program product for predicting settlement of high-rise buildings, in which the trend item, periodic item and random item are separated from multi-source monitoring data, and a nonlinear time series prediction model is constructed; the historical multi-source monitoring data is input into the nonlinear time series prediction model to obtain preliminary settlement prediction results; a continuous settlement distribution map of the high-rise building is generated, and the spatial distribution characteristics are obtained; a number of key influencing factors and corresponding influencing factor weights are determined based on the multi-source monitoring data; a settlement prediction model is constructed; the multi-source monitoring data is input into the settlement prediction model to obtain the final settlement prediction results; and the settlement trend curve and prediction warning time points are sent to the management terminal. The present application realizes early settlement prediction of high-rise buildings, improves the timeliness of detecting settlement risks in high-rise buildings, and reduces the probability of safety hazards caused by settlement of high-rise buildings.

Description

Technical Field

[0001] The present application belongs to the field of measuring distance, level or orientation, and in particular to a high-rise building settlement prediction method, system, storage medium and program product. Background Art

[0002] The safety and stability of high-rise buildings have always been a key concern of the engineering community, and building settlement monitoring is a key link in ensuring the safety of high-rise buildings. With the acceleration of urbanization and the increase in the number of high-rise buildings, settlement monitoring technology is also constantly developing and improving. Although traditional manual measurement methods are reliable, they face challenges in efficiency and real-time performance when used on a large scale.

[0003] In recent years, automated settlement monitoring systems have been widely used and developed. In related technologies, multiple sensors are usually installed at key locations of buildings, such as tilt sensors, displacement sensors, and strain sensors. These sensors can continuously collect settlement data of buildings and transmit it to the central monitoring system in real time through wireless communication networks. The monitoring system uses specialized data processing software to analyze and visualize the collected data, realizing real-time monitoring of building settlement conditions and automatic alarm.

[0004] Nevertheless, since the settlement process of high-rise buildings is affected by many factors and presents complex temporal and spatial variation characteristics, it may be difficult to fully capture the dynamic change process of settlement by relying solely on traditional sensor real-time data judgment, especially in the early stages, when the settlement trend is not obvious and it is difficult to identify potential settlement risks. Summary of the invention

[0005] The present application provides a high-rise building settlement prediction method, system, storage medium and program product, which are used to realize early settlement prediction of high-rise buildings, generate settlement trend curves and prediction warning time points based on real-time data from sensors, and send the final prediction results to the management terminal so that relevant personnel can make response plans in advance, thereby improving the timeliness of detecting settlement risks in high-rise buildings and reducing the probability of safety hazards caused by settlement of high-rise buildings.

[0006] In the first aspect, the present application provides a high-rise building settlement prediction method, which uses an empirical mode decomposition method to separate trend terms, periodic terms and random terms from multi-source monitoring data detected by a number of preset sensors, and constructs a nonlinear time series prediction model based on the trend terms, periodic terms and random terms;

[0007] Input historical multi-source monitoring data into the nonlinear time series prediction model to obtain preliminary settlement prediction results; generate a continuous settlement distribution map of high-rise buildings based on the spatial information in the multi-source monitoring data, and obtain spatial distribution characteristics based on the continuous settlement distribution map; determine several key influencing factors and their corresponding influencing factor weights based on the multi-source monitoring data;

[0008] A settlement prediction model is constructed based on preliminary settlement prediction results, spatial distribution characteristics and influencing factor weights. The settlement prediction model is a model obtained by inputting the preliminary settlement prediction results, spatial distribution characteristics and influencing factor weights as a training set into a preset neural network for training. The monitoring data obtained by the preset sensor detection is input into the settlement prediction model, and the settlement prediction model outputs the prediction results.

[0009] Input the multi-source monitoring data into the settlement prediction model to obtain the final settlement prediction results, which include the settlement trend curve and the prediction and warning time points;

[0010] The settlement trend curve and predicted warning time points are sent to the management terminal.

[0011] By adopting the above technical solution, the empirical mode decomposition method is used to separate different feature items from multi-source monitoring data. The system can identify and analyze various change patterns in building settlement data, thereby extracting more accurate settlement characteristics. The nonlinear time series prediction model constructed by combining trend terms, periodic terms and random terms can fully reflect the dynamic characteristics of the settlement process. On this basis, the continuous settlement distribution map generated by the system using spatial information reflects the overall settlement state of the building. The key influencing factors and their weights determined by multi-source monitoring data improve the model's ability to describe the settlement mechanism. These elements are integrated into the settlement prediction model trained in the preset neural network, so that the prediction results simultaneously consider the comprehensive effects of time series characteristics, spatial distribution characteristics and influencing factors. The settlement trend curve and prediction warning time point obtained in this way have higher prediction accuracy and can provide more reliable decision-making basis for building managers. It realizes early settlement prediction of high-rise buildings, generates settlement trend curves and prediction warning time points based on real-time data from sensors, and sends the final prediction results to the management terminal so that relevant personnel can make response plans in advance, thereby improving the timeliness of detecting settlement risks in high-rise buildings and reducing the probability of safety hazards caused by settlement in high-rise buildings.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the multi-source monitoring data detected by a plurality of preset sensors is separated into trend items, period items and random items using an empirical mode decomposition method, specifically including:

[0013] Adopting adaptive wavelet threshold denoising algorithm to preprocess multi-source monitoring data, and obtaining preprocessed multi-source monitoring data;

[0014] Combining variational mode decomposition and empirical mode decomposition algorithms, a hybrid decomposition model is constructed;

[0015] The information entropy criterion is used to determine the optimal number of decomposition layers, and the hybrid decomposition model is applied to decompose the preprocessed multi-source monitoring data to obtain the components and their corresponding time-frequency characteristics;

[0016] Based on the time-frequency characteristics, a machine learning classification algorithm is used to classify the components into trend items, periodic items, and random items.

[0017] By adopting the above technical solution, the multi-source monitoring data is preprocessed with an adaptive wavelet threshold denoising algorithm, which reduces the noise interference in the original data. By combining the variational mode decomposition and empirical mode decomposition algorithms to construct a hybrid decomposition model, the limitations of the single decomposition method are reduced and the accuracy of the decomposition results is improved. The optimal number of decomposition layers is determined using the information entropy criterion, which reduces the adverse effects of over-decomposition or insufficient decomposition, so that the decomposed components better reflect the essential characteristics of the data. Based on the time-frequency characteristics, a machine learning classification algorithm is used to classify the components into trend items, periodic items, and random items, which realizes the accurate identification of the components of the sedimentation data, so that the subsequent prediction model can adopt more targeted processing methods for components with different characteristics, thereby improving the overall prediction accuracy.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, the step of generating a continuous settlement distribution map of a high-rise building based on spatial information in multi-source monitoring data specifically includes:

[0019] Generate initial continuous settlement distribution map based on interpolation algorithm from multi-source monitoring data;

[0020] Combined with the three-dimensional structural information of high-rise buildings, a settlement distribution model with physical constraints is constructed;

[0021] The variational Bayesian inference method is used to fuse the initial continuous settlement distribution map and the settlement distribution model to generate the continuous settlement distribution map of the high-rise building.

[0022] By adopting the above technical solution, the initial continuous settlement distribution map is generated based on the multi-source monitoring data using the interpolation algorithm, which solves the problem of spatial information discontinuity caused by the discrete distribution of monitoring points. By combining the three-dimensional structural information of high-rise buildings to construct a settlement distribution model with physical constraints, the generated settlement distribution conforms to the actual structural characteristics and mechanical properties of the building. The variational Bayesian inference method is used to fuse the initial continuous settlement distribution map and the settlement distribution model, while maintaining the statistical characteristics of the data and satisfying the physical constraints. The generated high-rise building continuous settlement distribution map not only reflects the characteristics of the measured data, but also conforms to the settlement deformation law of the building, providing more accurate spatial distribution feature information for subsequent predictions.

[0023] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of constructing a settlement prediction model based on the preliminary settlement prediction results, the spatial distribution characteristics and the weights of the influencing factors, the method further includes:

[0024] Construct a multi-task learning framework that can predict the amount of sedimentation, sedimentation rate, and sedimentation acceleration;

[0025] Capturing the dynamic spatial dependencies between several preset sensors based on a multi-task learning framework;

[0026] The settlement prediction model is dynamically adjusted based on dynamic spatial dependencies and the Bayesian optimization algorithm is applied.

[0027] By adopting the above technical solutions, the constructed multi-task learning framework can achieve a comprehensive characterization of the building settlement state by simultaneously predicting the settlement amount, settlement rate and settlement acceleration. Based on the multi-task learning framework, the dynamic spatial dependency between the preset sensors is captured, reflecting the mutual influence of settlement changes between the monitoring points of the building. The settlement prediction model is dynamically adjusted through the Bayesian optimization algorithm so that the model can adapt to the dynamic change characteristics of the building settlement process. The adaptability of the model to complex settlement processes is improved, so that the prediction results can better reflect the overall settlement and deformation laws of the building, and enhance the generalization performance of the prediction model.

[0028] In combination with some embodiments of the first aspect, in some embodiments, capturing the dynamic spatial dependency relationship between a plurality of preset sensors based on a multi-task learning framework specifically includes:

[0029] A time-varying graph structure is constructed, and several preset sensors are represented as nodes in the time-varying graph structure, and dynamic spatial dependencies are represented as edges in the time-varying graph structure;

[0030] Determine the dependency between each node based on the time-varying graph structure;

[0031] Capture the temporal evolution of dependencies based on the time-aware attention mechanism;

[0032] Based on the multi-task learning framework, the dynamic spatial dependency relationship between several preset sensors is established according to the time evolution characteristics.

[0033] By adopting the above technical solution, a time-varying graph structure is constructed and the preset sensors are represented as nodes, and the dynamic spatial dependencies are represented as edges, thus realizing the mathematical expression of the complex relationship between the monitoring points. The dependencies between nodes determined according to the time-varying graph structure reflect the interaction law of the settlement changes of different monitoring points. The time-aware attention mechanism is used to capture the time evolution characteristics of the dependencies, so that the system can identify and track the dynamic changes of the relationship between the monitoring points over time. The dynamic spatial dependency relationship established based on the time evolution characteristics based on the multi-task learning framework not only reflects the spatial correlation between the monitoring points, but also includes the changing characteristics of the correlation over time, which improves the model's ability to express the overall settlement behavior of the building, enabling the prediction model to more accurately characterize the mutual influence and evolution law between the monitoring points, thereby improving the accuracy of settlement prediction.

[0034] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of inputting the multi-source monitoring data into the settlement prediction model to obtain the final settlement prediction result, the method further includes:

[0035] Quantify the uncertainty of the final settlement prediction results and generate confidence intervals and probability distributions of the prediction values;

[0036] Build multi-scale forecasting models that include short-term, medium-term, and long-term forecasts;

[0037] Set several groups of preset external condition parameters, use the settlement prediction model to perform scenario simulation, and obtain the settlement development trends corresponding to the several groups of preset external condition parameters;

[0038] Determine a matching settlement development trend having the highest similarity to the settlement trend curve graph and a corresponding matching preset external condition parameter among several settlement development trends;

[0039] Based on matching the preset external condition parameters and the final settlement prediction results, the weighted sliding time window method is used to dynamically calculate and update the warning time point;

[0040] The final settlement prediction results, matching settlement development trends and updated warning time points are fused with the building information model of the high-rise building to generate a four-dimensional visualization model, in which time is used as the fourth dimension.

[0041] By adopting the above technical solutions, the confidence interval and probability distribution obtained by quantifying the uncertainty of the final settlement prediction results provide a reliability assessment for the prediction results. By constructing a multi-scale prediction model that includes short-term, medium-term and long-term predictions, the prediction results can meet the monitoring needs of different time scales. The settlement prediction model is used to simulate different external condition parameters, and a variety of possible settlement development trends are obtained. The most matching development trend and external condition parameters are determined by similarity comparison. The weighted sliding time window method is used to dynamically calculate and update the warning time point, which improves the accuracy and real-time nature of the warning time point. The four-dimensional visualization model generated by integrating the prediction results with the building information model realizes the intuitive display of the settlement prediction results, enabling building managers to better understand and apply the prediction results, and improving the practicality of the prediction results.

[0042] In combination with some embodiments of the first aspect, in some embodiments, after the step of sending the settlement trend curve graph and the prediction and warning time points to the management terminal, the method further includes:

[0043] Based on the final settlement prediction results, search for the corresponding settlement risk level in the preset database;

[0044] The corresponding preset warning plan is matched according to the settlement risk level, and the preset warning plan is sent to the management terminal.

[0045] By adopting the above technical solution, the corresponding settlement risk level is searched in the preset database based on the final settlement prediction result, and a direct connection between the prediction result and the risk assessment is established. The preset early warning plan is matched according to the settlement risk level and sent to the management terminal, so that the prediction results can directly guide the formulation and implementation of actual prevention and control measures, improving the timeliness and effectiveness of building settlement risk prevention and control.

[0046] In a second aspect, an embodiment of the present application provides a high-rise building settlement prediction system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0047] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a system, causes the system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0048] In a fourth aspect, an embodiment of the present application provides a computer program product, characterized in that when the computer program product runs on a system, the system executes the method described in any possible implementation manner in the first aspect.

[0049] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0050] 1. The present application provides a method for predicting the settlement of high-rise buildings. The empirical mode decomposition method is used to separate different feature items from multi-source monitoring data. The system can identify and analyze various change patterns in building settlement data, thereby extracting more accurate settlement features. The nonlinear time series prediction model constructed by combining trend terms, periodic terms and random terms can fully reflect the dynamic characteristics of the settlement process. On this basis, the continuous settlement distribution map generated by the system using spatial information reflects the overall settlement state of the building. The key influencing factors and their weights determined by multi-source monitoring data improve the model's ability to characterize the settlement mechanism. These elements are integrated into the settlement prediction model trained in the preset neural network, so that the prediction results simultaneously consider the comprehensive effects of time series characteristics, spatial distribution characteristics and influencing factors. The settlement trend curve and prediction warning time points obtained in this way have higher prediction accuracy and can provide more reliable decision-making basis for building managers. It realizes early settlement prediction of high-rise buildings, generates settlement trend curves and prediction warning time points based on real-time data from sensors, and sends the final prediction results to the management terminal so that relevant personnel can make response plans in advance, thereby improving the timeliness of detecting settlement risks in high-rise buildings and reducing the probability of safety hazards caused by settlement in high-rise buildings.

[0051] 2. This application provides a method for predicting the settlement of high-rise buildings. The constructed multi-task learning framework achieves a comprehensive characterization of the settlement state of the building by simultaneously predicting the settlement amount, settlement rate and settlement acceleration. Based on the multi-task learning framework, the dynamic spatial dependency between preset sensors is captured, reflecting the mutual influence of settlement changes between monitoring points of the building. The settlement prediction model is dynamically adjusted through the Bayesian optimization algorithm so that the model can adapt to the dynamic change characteristics during the building settlement process. The adaptability of the model to complex settlement processes is improved, so that the prediction results can better reflect the overall settlement and deformation laws of the building, and enhance the generalization performance of the prediction model.

[0052] 3. The present application provides a method for predicting the settlement of high-rise buildings, and the confidence interval and probability distribution obtained by quantifying the uncertainty of the final settlement prediction results provide a reliability assessment for the prediction results. By constructing a multi-scale prediction model that includes short-term, medium-term and long-term predictions, the prediction results can meet the monitoring needs of different time scales. The settlement prediction model is used to simulate different external condition parameters, and a variety of possible settlement development trends are obtained, and the most matching development trends and external condition parameters are determined by similarity comparison. The weighted sliding time window method is used to dynamically calculate and update the warning time point, which improves the accuracy and real-time nature of the warning time point. The four-dimensional visualization model generated by integrating the prediction results with the building information model realizes the intuitive display of the settlement prediction results, enabling building managers to better understand and apply the prediction results, thereby improving the practicality of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of a high-rise building settlement prediction method in an embodiment of the present application.

[0054] Figure 2 It is a flow chart of a method for supplementing the final settlement prediction result in an embodiment of the present application.

[0055] Figure 3 It is a schematic diagram of the physical device structure of a high-rise building settlement prediction system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more listed items.

[0057] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0058] Combine the following Figure 1 , a high-rise building settlement prediction method in an embodiment of the present application is described:

[0059] See also Figure 1 , which is a flow chart of a high-rise building settlement prediction method in an embodiment of the present application.

[0060] S101, using an empirical mode decomposition method to separate trend items, period items, and random items from multi-source monitoring data detected by a number of preset sensors, and constructing a nonlinear time series prediction model based on the trend items, period items, and random items;

[0061] The system uses the empirical mode decomposition method to separate the trend term, periodic term and random term of the multi-source monitoring data detected by several preset sensors. Specifically: the system uses the adaptive wavelet threshold denoising algorithm to pre-process the multi-source monitoring data to obtain the pre-processed multi-source monitoring data;

[0062] Combining variational mode decomposition and empirical mode decomposition algorithms, a hybrid decomposition model is constructed;

[0063] The information entropy criterion is used to determine the optimal number of decomposition layers, and the hybrid decomposition model is applied to decompose the preprocessed multi-source monitoring data to obtain the components and their corresponding time-frequency characteristics;

[0064] Based on the time-frequency characteristics, a machine learning classification algorithm is used to classify the components into trend items, periodic items, and random items.

[0065] The system then constructs a nonlinear time series prediction model based on trend terms, cycle terms, and random terms.

[0066] The system first uses an adaptive wavelet threshold denoising algorithm to preprocess the multi-source monitoring data. Specifically, it decomposes the original data by selecting appropriate wavelet basis functions, and sets adaptive thresholds at different scales for noise reduction. On this basis, the system combines variational mode decomposition and empirical mode decomposition algorithms to construct a hybrid decomposition model, in which variational mode decomposition decomposes the signal into a series of intrinsic mode functions through iterative optimization, and empirical mode decomposition obtains intrinsic mode functions of different scales through envelope screening methods. The system calculates the amount of information at different decomposition levels based on the information entropy criterion, and selects the number of levels with the optimal information entropy as the final decomposition level. For each component obtained by decomposition, the system extracts time domain and frequency domain feature parameters, including energy distribution, frequency centrality, bandwidth and other features, and identifies these components as trend terms, periodic terms and random terms through machine learning classification algorithms, and finally constructs a nonlinear time series prediction model.

[0067] Furthermore, the system can achieve refined analysis and prediction of building settlement data by constructing a multi-level decomposition framework, introducing singular spectrum analysis methods, establishing a deep learning feature extraction mechanism, and adopting a hybrid integrated prediction strategy. Specifically, the system can construct a multi-level decomposition framework in which different decomposition strategies are adopted for signals in different frequency bands. For example, the system can use wavelet packet decomposition methods to extract local features in the high frequency band, use variational mode decomposition to extract periodic features in the medium frequency band, and use empirical mode decomposition to extract trend features in the low frequency band. The system can also introduce singular spectrum analysis methods to achieve refined extraction of signal components through trajectory matrix decomposition and reconstruction. In the classification stage, the system can establish a deep learning-based automatic feature extraction mechanism to improve the accuracy of component identification through multi-level feature learning. For the construction of the prediction model, the system can adopt a hybrid integration strategy to adaptively combine different types of prediction models.

[0068] S102, inputting historical multi-source monitoring data into a nonlinear time series prediction model to obtain preliminary settlement prediction results; generating a continuous settlement distribution map of high-rise buildings based on spatial information in the multi-source monitoring data, and obtaining spatial distribution characteristics based on the continuous settlement distribution map; determining a number of key influencing factors and corresponding influencing factor weights based on the multi-source monitoring data;

[0069] The system builds a prediction model based on the trend term, periodic term and random term components decomposed in the previous step. The trend term prediction model predicts long-term trend changes. The trend term prediction model uses the polynomial regression method to fit historical data. The system determines the optimal order of the polynomial based on the goodness of fit. The system uses the Fourier analysis method for the periodic term, determines the main periodic components through spectrum analysis, and establishes a periodic term prediction model. For the random term, the system establishes an autoregressive moving average model and determines the model order by analyzing the autocorrelation and partial autocorrelation of the random term. Finally, the system combines the prediction results of the three sub-models to obtain the overall prediction result of building settlement.

[0070] Based on this prediction step, the system can also achieve dynamic and accurate prediction of building settlement by constructing a dynamic prediction framework and introducing kernel regression method, wavelet time-frequency analysis and conditional heteroscedasticity model in each sub-model. Specifically, the system can build a dynamic prediction framework, which contains an online update mechanism for the prediction model. Specifically, the system continuously updates the training data set and re-estimates the model parameters through the sliding time window method. The system introduces the kernel regression method in the trend term prediction and realizes accurate modeling of nonlinear trends through local weighting. For periodic terms, the system introduces wavelet transform for time-frequency joint analysis to capture non-stationary periodic characteristics. In the prediction of random terms, the system uses the conditional heteroscedasticity model to characterize the fluctuation characteristics.

[0071] S103, constructing a settlement prediction model based on preliminary settlement prediction results, spatial distribution characteristics and influencing factor weights;

[0072] A settlement prediction model is constructed based on preliminary settlement prediction results, spatial distribution characteristics and influencing factor weights. The settlement prediction model is a model obtained by inputting preliminary settlement prediction results, spatial distribution characteristics and influencing factor weights as a training set into a preset neural network for training. The monitoring data obtained by the preset sensor detection is input into the settlement prediction model, and the settlement prediction model outputs the prediction results.

[0073] The system constructs a settlement prediction model based on preliminary settlement prediction results, spatial distribution characteristics and influencing factor weights, and uses multi-layer perception and neural network as the basic architecture. The input layer of the neural network contains three branches: the first branch receives preliminary settlement prediction results, including trend prediction values, period prediction values ​​and random prediction values; the second branch receives spatial distribution characteristics, including settlement contour parameters, settlement gradient parameters and settlement curvature parameters; the third branch receives influencing factor characteristics, including the normalized values ​​of each key influencing factor and its corresponding weight. The system adopts a multi-layer structure design in the hidden layer of the neural network. Each layer uses the ReLU activation function to introduce nonlinear characteristics, and the normalization layer is used to reduce the internal covariate offset. The system introduces a residual connection mechanism between network layers to effectively prevent the gradient disappearance problem during deep network training. The system uses dropout technology to randomly disconnect some neuron connections to enhance the generalization ability of the model. The system uses the Adam optimizer to train the neural network, improves the training efficiency by dynamically adjusting the learning rate, and adopts an early stopping strategy to prevent overfitting. The system designs multiple output nodes in the output layer, corresponding to settlement prediction values ​​of different time scales.

[0074] Based on the model construction steps, the system can achieve high-precision prediction modeling of building settlement by introducing attention mechanism, gated recurrent unit, bidirectional propagation structure, integrated learning strategy and transfer learning mechanism. Specifically: the system can use attention mechanism to enhance the expressiveness of features, and adaptively weight different features through self-attention module. The system introduces gated recurrent unit to process time series features and captures long-term dependencies through memory mechanism. The system constructs a bidirectional propagation structure to extract feature information simultaneously through forward and backward propagation. The system adopts integrated learning strategy to improve the stability of prediction through model fusion. The system introduces transfer learning mechanism and accelerates model convergence by using pre-trained model.

[0075] S104, inputting the multi-source monitoring data into the settlement prediction model to obtain the final settlement prediction result;

[0076] The multi-source monitoring data is input into the settlement prediction model to obtain the final settlement prediction result, which includes the settlement trend curve and prediction warning time points, and the settlement trend curve and prediction warning time points are sent to the management terminal.

[0077] The system inputs the multi-source monitoring data into the trained settlement prediction model to obtain the final settlement prediction results. These multi-source monitoring data include but are not limited to real-time data from displacement sensors, pressure gauges, inclinometers, etc., which reflect the current physical state of the building and changes in the surrounding environment. The settlement prediction model comprehensively considers the various input factors in step S103, and outputs a prediction of the settlement of the building in the future through the complex calculation process of the neural network, including a settlement trend curve and a prediction warning time point. These prediction results are then sent to the management terminal for engineers and decision makers to evaluate and make further decisions.

[0078] The system can enhance the adaptability and real-time response capability of the prediction model by implementing real-time data analysis and dynamic model updating. For example, the introduction of online learning mechanisms enables the model to be updated instantly as new data is continuously input, thereby more accurately reflecting the impact of real-time environmental and structural changes. It is also possible to deploy distributed sensor networks to achieve more extensive data collection and enhance the quality and coverage of the model's input data.

[0079] S105. Build a multi-task learning framework;

[0080] The system constructs a multi-task learning framework, which is a framework that can predict the amount of settlement, settlement rate and settlement acceleration.

[0081] The system builds a multi-task learning framework that can simultaneously predict settlement, settlement rate, and settlement acceleration. Multi-task learning is a machine learning method that simultaneously learns multiple related tasks in one model to improve the learning effect and prediction accuracy of each task. This method can optimize model parameters and reduce the risk of overfitting by taking advantage of the common features between different tasks. In this step, each task not only independently predicts its own target value, but also improves the overall learning efficiency and prediction accuracy by sharing hidden layers. The system uses this framework to conduct a comprehensive analysis of different aspects of building settlement and achieve comprehensive monitoring of the building's health.

[0082] The system can comprehensively improve the effectiveness and application scope of the multi-task learning framework by adding relevant prediction tasks and introducing advanced learning algorithms. Specifically, the system further improves the performance of the multi-task learning framework by introducing more advanced deep learning algorithms such as deep belief networks (DBN). In addition, by adding more relevant prediction tasks, such as foundation change analysis and surrounding building impact assessment, the learning framework can be made more comprehensive, thereby improving the comprehensiveness and accuracy of the prediction. The system can also use transfer learning technology to apply the learning results of existing buildings to the settlement prediction tasks of new buildings, thereby reducing the cost and time of model training for new projects.

[0083] S106, capturing dynamic spatial dependencies between a number of preset sensors based on a multi-task learning framework;

[0084] The system captures the dynamic spatial dependency between several preset sensors based on a multi-task learning framework. Specifically: a time-varying graph structure is constructed, and several preset sensors are represented as nodes in the time-varying graph structure, and the dynamic spatial dependency is represented as edges in the time-varying graph structure;

[0085] Determine the dependency between each node based on the time-varying graph structure;

[0086] Capture the temporal evolution of dependencies based on the time-aware attention mechanism;

[0087] Based on the multi-task learning framework, the dynamic spatial dependency relationship between several preset sensors is established according to the time evolution characteristics.

[0088] The system captures the dynamic spatial dependencies between several preset sensors based on a multi-task learning framework. The system first constructs a time-varying graph structure, represents the sensors as nodes in the graph, and the connections between the nodes are represented as dynamic spatial dependencies. This structure allows the system to monitor and analyze the interactions between sensors and their dynamic characteristics over time. In this way, the system can adjust the monitoring strategy of the building structure in real time and optimize the data collection and processing process. The analysis of dynamic spatial dependencies not only improves the efficiency of data utilization, but also enhances the adaptability of the prediction model to environmental changes.

[0089] The system can process data in time-varying graph structures by introducing more complex graph network algorithms, such as graph convolutional networks (GCNs), so that the model can have a deeper understanding of the complex dependencies between nodes. In addition, the system can develop new algorithms to optimize the dynamic update mechanism of the graph structure and ensure the real-time accuracy of nodes and edges to adapt to the rapidly changing building environment. Through the application of these technologies, the system not only improves the efficiency of data processing, but also ensures the timeliness and accuracy of the model output.

[0090] S107. Dynamically adjust the settlement prediction model based on dynamic spatial dependency and by applying a Bayesian optimization algorithm.

[0091] The system dynamically adjusts the settlement prediction model based on dynamic spatial dependencies and applies the Bayesian optimization algorithm. Bayesian optimization is an optimization technique based on a probabilistic model that is used to optimize a computationally expensive objective function. In this way, the system continuously updates and adjusts the parameters of the settlement prediction model to respond to changes in input data or environmental conditions. This dynamic adjustment process keeps the model in an optimal state and accurately reflects the actual settlement of the building.

[0092] The system can combine Bayesian optimization with other parameter optimization techniques, such as genetic algorithms or particle swarm optimization, to further improve the efficiency and effectiveness of parameter adjustment. In addition, the system can develop an adaptive learning algorithm that enables the model to automatically select the most appropriate optimization strategy at different monitoring stages, thereby achieving more refined and personalized model adjustment.

[0093] In the above embodiment, the empirical mode decomposition method is used to separate different feature items from multi-source monitoring data, and the system can identify and analyze various change patterns in building settlement data, thereby extracting more accurate settlement features. The nonlinear time series prediction model constructed by combining trend terms, periodic terms and random terms can fully reflect the dynamic characteristics of the settlement process. On this basis, the continuous settlement distribution map generated by the system using spatial information reflects the overall settlement state of the building, and the key influencing factors and their weights determined by multi-source monitoring data improve the model's ability to characterize the settlement mechanism. These elements are integrated into the settlement prediction model obtained by training in the preset neural network, so that the prediction results simultaneously consider the comprehensive effects of time series characteristics, spatial distribution characteristics and influencing factors. The settlement trend curve and prediction warning time points obtained in this way have higher prediction accuracy and can provide more reliable decision-making basis for building management personnel. It realizes early settlement prediction of high-rise buildings, generates settlement trend curves and prediction warning time points based on real-time data from sensors, and sends the final prediction results to the management terminal so that relevant personnel can make response plans in advance, thereby improving the timeliness of detecting settlement risks in high-rise buildings and reducing the probability of safety hazards caused by settlement in high-rise buildings.

[0094] After the final settlement prediction result is obtained in the above embodiment, it is necessary to further improve the information sent to the management terminal. The information needs to contain a variety of data so that relevant staff can quickly and clearly understand the prediction result. Figure 2 , a method for supplementing the final settlement prediction result in the embodiment of the present application is described:

[0095] See also Figure 2, which is a flow chart of a method for supplementing the final settlement prediction result in an embodiment of the present application.

[0096] S201, quantifying the uncertainty of the final settlement prediction result, and generating a confidence interval and probability distribution of the prediction value;

[0097] The system first performs a statistical analysis on the settlement monitoring data to determine the uncertainty of the final settlement prediction results. This step is achieved by calculating the confidence interval and probability distribution of the predicted value. The confidence interval provides the credibility of the prediction results, while the probability distribution reveals the probability of each possible value of the prediction result, which is critical for subsequent risk assessment and decision making. The system can use historical settlement data, combined with the actual situation of the current building and external environmental factors, and apply statistical methods such as Monte Carlo simulation or Bayesian inference to derive the uncertainty range of settlement prediction.

[0098] The system can further develop customized uncertainty assessment models for different building types and geological conditions. For example, for buildings located in geologically unstable areas, the system can combine geological scanning data and historical settlement records to use more complex geological dynamic simulation technology to refine the prediction of uncertainty. In addition, the system can also combine the uncertainty analysis results with the building's use function and importance level to develop a more detailed risk management strategy.

[0099] S202, construct a multi-scale prediction model including short-term, medium-term and long-term predictions;

[0100] The system responds to settlement prediction needs at different time scales by building a multi-scale prediction model that includes short-term (e.g., within one month), medium-term (within one year), and long-term (ten years and above) predictions. This model can provide appropriate prediction accuracy and detail levels according to different time frames, helping decision makers to conduct more accurate risk assessments and resource allocation. Multi-scale models usually need to take into account a variety of factors, such as historical settlement data, building loads, changes in soil properties, groundwater level changes, etc., all of which may affect the settlement of buildings at different time scales.

[0101] Based on the multi-scale prediction model, the system can develop an adaptive prediction model adjustment mechanism to cope with possible external environmental changes, such as extreme weather events, abnormal changes in groundwater levels, etc. This mechanism can monitor environmental data in real time and dynamically adjust the parameters of the prediction model according to data changes to maintain the accuracy of the prediction. In addition, the system can also combine this multi-scale model with real-time data analysis technology to realize a real-time early warning system and enhance the real-time and predictive nature of building safety management.

[0102] S203, setting several groups of preset external condition parameters, using the settlement prediction model to perform scenario simulation, and obtaining settlement development trends corresponding to the several groups of preset external condition parameters;

[0103] This step involves setting several sets of preset external condition parameters, which are used to simulate the settlement response of buildings under different environmental conditions. These external condition parameters may include geological conditions, climate change, rainfall, groundwater level changes, etc. Through these parameter settings, the system can evaluate the settlement trend of buildings under different external influences. This step is carried out through the settlement prediction model. By inputting different parameter combinations, a variety of possible settlement scenarios are simulated, which helps decision makers conduct a more comprehensive risk assessment.

[0104] After completing the basic external condition parameter settings, the system can further develop a flexible parameter adjustment and optimization tool that can automatically adjust the external condition parameters based on real-time monitoring data to reflect the latest environmental changes. This automatic adjustment mechanism can greatly improve the real-time and accuracy of the simulation, making the prediction results more in line with the actual situation. In addition, the system can also use artificial intelligence technology, such as machine learning, to analyze the relationship between past parameter settings and actual settlement results, optimize parameter setting strategies, and improve the accuracy of the prediction model.

[0105] S204, determining a matching settlement development trend having the highest similarity to the settlement trend curve graph and a corresponding matching preset external condition parameter among several settlement development trends;

[0106] The core of this step is to use image recognition or pattern matching technology to analyze and compare different settlement trend curves. The system first compares the settlement trend curves simulated under multiple preset external conditions with the actually observed settlement trend curves, and uses an algorithm to calculate the curve with the highest similarity. This process may include calculating multiple dimensions such as the Euclidean distance between the two curves, the contrast of the curve shapes, and the fit of the curves. In this way, the system can accurately select a prediction curve that is closest to the actual settlement situation, providing a basis for subsequent early warning and decision-making.

[0107] The system can further develop more advanced curve matching algorithms, such as automatically identifying and matching the closest development trends through deep learning training models. In addition, the system can be extended to other types of pattern recognition tasks, such as climate change trend prediction, trend analysis of economic data, etc. The system can improve the accuracy of matching and the versatility of the model by integrating more types of data inputs, such as geological data, historical settlement data, etc. Through this technical expansion, the system can not only improve the accuracy of predictions, but also be applied to a wider range of fields, thereby achieving more comprehensive risk management and decision support.

[0108] S205, based on matching preset external condition parameters and final settlement prediction results, dynamically calculate and update the warning time point using a weighted sliding time window method;

[0109] In this step, the system creates a three-dimensional visual representation that includes the time dimension by merging structured data (such as settlement prediction results) with unstructured data (such as building models). This allows stakeholders to intuitively see the structural changes that may occur in the building over time. The generation of the four-dimensional model relies on advanced graphics processing technology and data fusion algorithms to ensure seamless integration and real-time updates of various data sources.

[0110] S206, the final settlement prediction result, the matching settlement development trend and the updated warning time point are integrated with the building information model of the high-rise building to generate a four-dimensional visualization model.

[0111] The system fuses the final settlement prediction results, matching settlement development trends and updated warning time points with the building information model of the high-rise building to generate a four-dimensional visualization model, in which time is the fourth dimension.

[0112] Based on this step, the system can further develop its data fusion and visualization technology, such as through augmented reality (AR) or virtual reality (VR) technology, to provide users with a more immersive viewing and analysis experience. This technology can also be applied to fields such as urban planning and disaster management, helping experts and decision makers better understand complex data and trends through four-dimensional visualization.

[0113] In the above embodiment, the confidence interval and probability distribution obtained by quantifying the uncertainty of the final settlement prediction result provide a reliability assessment for the prediction result. By constructing a multi-scale prediction model including short-term, medium-term and long-term predictions, the prediction results can meet the monitoring needs of different time scales. The settlement prediction model is used to simulate different external condition parameters, and a variety of possible settlement development trends are obtained, and the most matching development trends and external condition parameters are determined by similarity comparison. The weighted sliding time window method is used to dynamically calculate and update the warning time point, which improves the accuracy and real-time nature of the warning time point. The four-dimensional visualization model generated by integrating the prediction results with the building information model realizes the intuitive display of the settlement prediction results, enabling building managers to better understand and apply the prediction results, thereby improving the practicality of the prediction results.

[0114] The following describes the system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of a high-rise building settlement prediction system provided in an embodiment of the present application.

[0115] It should be noted that Figure 3 The structure of the system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0116] like Figure 3 As shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303, such as executing the method in the above embodiment. In RAM 303, various programs and data required for system operation are also stored. CPU 301, ROM 302 and RAM 303 are connected to each other through a bus 304. Input / output (I / O) interface 305 is also connected to bus 304.

[0117] The following components are connected to the I / O interface 305: an input section 306 including a camera, an infrared sensor, etc.; an output section 307 including a liquid crystal display (LCD) and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0118] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 309, and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0119] It should be noted that the computer-readable medium shown in the embodiment of the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.

[0120] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Among them, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0121] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the system described in the above embodiment; or may exist independently without being assembled into the system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of a system, the system implements the method provided in the above embodiment.

[0122] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0123] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0124] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk), etc.

[0125] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A high-rise building settlement prediction method, characterized in that: include: Using an empirical mode decomposition method to separate trend items, period items and random items from multi-source monitoring data detected by a number of preset sensors, and constructing a nonlinear time series prediction model based on the trend items, the period items and the random items; Inputting historical multi-source monitoring data into the nonlinear time series prediction model to obtain preliminary settlement prediction results; generating a continuous settlement distribution map of the high-rise building based on the spatial information in the multi-source monitoring data, and obtaining spatial distribution characteristics according to the continuous settlement distribution map; Determine a number of key influencing factors and corresponding influencing factor weights based on the multi-source monitoring data; A settlement prediction model is constructed based on the preliminary settlement prediction results, the spatial distribution characteristics and the weights of the influencing factors. The settlement prediction model is a model obtained by inputting the preliminary settlement prediction results, the spatial distribution characteristics and the weights of the influencing factors as a training set into a preset neural network for training. The monitoring data obtained by the preset sensor detection is input into the settlement prediction model, and the settlement prediction model outputs a prediction result. Inputting the multi-source monitoring data into the settlement prediction model to obtain a final settlement prediction result, wherein the final settlement prediction result includes a settlement trend curve graph and a prediction and warning time point; The settlement trend curve graph and the prediction and warning time points are sent to a management terminal.

2. The method according to claim 1, characterized in that The method of using an empirical mode decomposition method to separate trend items, periodic items and random items from multi-source monitoring data detected by a plurality of preset sensors specifically includes: Adopting adaptive wavelet threshold denoising algorithm to preprocess multi-source monitoring data, and obtaining preprocessed multi-source monitoring data; Combining variational mode decomposition and empirical mode decomposition algorithms, a hybrid decomposition model is constructed; Determine the optimal number of decomposition layers by using the information entropy criterion, and apply the hybrid decomposition model to decompose the pre-processed multi-source monitoring data to obtain components and time-frequency features corresponding to the components; Based on the time-frequency characteristics, a machine learning classification algorithm is used to classify the components into trend items, periodic items and random items.

3. The method according to claim 1, characterized in that The step of generating a continuous settlement distribution map of a high-rise building based on the spatial information in the multi-source monitoring data specifically includes: Generating an initial continuous settlement distribution map based on the multi-source monitoring data using an interpolation algorithm; Combined with the three-dimensional structural information of high-rise buildings, a settlement distribution model with physical constraints is constructed; The variational Bayesian inference method is used to fuse the initial continuous settlement distribution map and the settlement distribution model to generate a continuous settlement distribution map of the high-rise building.

4. The method according to claim 1, characterized in that: After the step of constructing a settlement prediction model based on the preliminary settlement prediction result, the spatial distribution characteristics and the weights of the influencing factors, the method further includes: Constructing a multi-task learning framework, wherein the multi-task learning framework is a framework that can predict the amount of sedimentation, the sedimentation rate, and the sedimentation acceleration; Capturing the dynamic spatial dependency between the plurality of preset sensors based on the multi-task learning framework; The settlement prediction model is dynamically adjusted based on the dynamic spatial dependency and by applying a Bayesian optimization algorithm.

5. The method according to claim 4, characterized in that The capturing of the dynamic spatial dependency relationship between the plurality of preset sensors based on the multi-task learning framework specifically includes: Constructing a time-varying graph structure, and representing the plurality of preset sensors as nodes in the time-varying graph structure, and representing the dynamic spatial dependency relationship as edges in the time-varying graph structure; Determine the dependency relationship between each of the nodes according to the time-varying graph structure; Capturing the temporal evolution characteristics of the dependencies based on a time-aware attention mechanism; Based on the multi-task learning framework, a dynamic spatial dependency relationship between the plurality of preset sensors is established according to the time evolution characteristics.

6. The method according to claim 1, characterized in that After the step of inputting the multi-source monitoring data into the settlement prediction model to obtain a final settlement prediction result, the method further includes: quantifying the uncertainty of the final settlement prediction result and generating a confidence interval and probability distribution of the prediction value; Build multi-scale forecasting models that include short-term, medium-term, and long-term forecasts; Setting a plurality of sets of preset external condition parameters, using the settlement prediction model to perform scenario simulation, and obtaining settlement development trends corresponding to the plurality of sets of preset external condition parameters; Determine, among the several settlement development trends, a matching settlement development trend having the highest similarity to the settlement trend curve graph and a corresponding matching preset external condition parameter; Based on the matching preset external condition parameters and the final settlement prediction result, the weighted sliding time window method is used to dynamically calculate and update the warning time point; The final settlement prediction result, the matching settlement development trend and the updated warning time point are data-fused with the building information model of the high-rise building to generate a four-dimensional visualization model, in which time is used as the fourth dimension.

7. The method according to claim 1, characterized in that After the step of sending the settlement trend curve graph and the prediction and warning time points to the management terminal, the method further includes: Based on the final settlement prediction result, searching for a corresponding settlement risk level in a preset database; A corresponding preset early warning plan is matched according to the subsidence risk level, and the preset early warning plan is sent to the management terminal.

8. A high-rise building settlement prediction system, characterized in that: The system comprises: One or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the system to execute the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to execute the method according to any one of claims 1 to 7.

Images