A method and system for predicting displacement trend of large-span spatial structures
Through the combination of vector autoregression and polynomial regression models, combining environment, load and structural characteristics, the displacement trend of large-span spatial structures is predicted, which solves the problem that traditional methods cannot capture displacement behavior and improves the safety and reliability of the structure.
Patent Information
- Application Number
- CN202411186451.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-08-28
AI Technical Summary
Traditional static analysis methods cannot fully capture the displacement behavior of large-span spatial structures, making it difficult to guarantee its safety and reliability during use.
Using a combination of vector autoregressive prediction model and polynomial regression prediction model, the environmental, load and structural feature data are extracted by acquiring and preprocessing monitoring data, and the prediction model is constructed to predict the displacement trend of large-span spatial structures, and the safety level evaluation is carried out. Early warning processing.
It improves the accuracy and reliability of large-span spatial structure displacement prediction, enhances the real-time monitoring capabilities of the structure, promptly detects potential anomalies, and improves safety and use reliability.
Smart Images

Figure CN118709307B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of large-span spatial structure displacement in civil engineering, and in particular to a large-span spatial structure displacement trend prediction method and system. Background Art
[0002] The large-span structure in civil engineering refers to a building structure with a larger span, which has unique design and construction requirements. The large-span structure can create an open column-free space, providing greater flexibility and visual transparency. Common examples include large stadiums and high-speed rail stations. These structures usually have high complexity and bearing capacity. However, while the large-span structure has many advantages, its construction has high requirements on the strength and performance of the materials. Due to its large span and complex structural form, the structural stress and deformation will be more significant. Traditional static analysis methods cannot fully capture its displacement behavior. With the development of current information technology and big data technology, precise calculations and advanced construction technologies have been used in large-span structures that are difficult to design and construct.
[0003] Therefore, the problem of how to predict the displacement trend of large-span space structures during use through effective methods and systems to improve the safety and reliability of large-span space structures needs to be solved urgently. Summary of the invention
[0004] The main purpose of this application is to provide a method and system for predicting the displacement trend of a large-span spatial structure, aiming to improve the reliability and accuracy of fault diagnosis of the displacement of a large-span spatial structure.
[0005] In order to achieve the above-mentioned invention object, the present application proposes a method for predicting displacement trend of a large-span spatial structure, the method comprising:
[0006] Acquire monitoring data of the large-span spatial structure, and pre-process the monitoring data to obtain first monitoring data;
[0007] Extracting features from the first monitoring data to obtain environmental feature data, load feature data, and structural feature data;
[0008] Pre-constructing a vector autoregression prediction model, and inputting the environmental characteristic data, the load characteristic data, and the structural characteristic data into the vector autoregression prediction model respectively, so that the vector autoregression prediction model outputs the environmental characteristic prediction data, the load characteristic prediction data, and the structural characteristic prediction data respectively;
[0009] A polynomial regression prediction model is constructed, and the environmental characteristic prediction data, the load characteristic prediction data and the structural characteristic prediction data are input into the polynomial regression prediction model to obtain a large-span spatial structure displacement trend prediction result.
[0010] In the above steps, feature extraction is performed on the first monitoring data to obtain load feature data, which specifically includes the following steps:
[0011] Extracting features from the first monitoring data to obtain static load data, dynamic load data, and variable load data; performing weighted sum processing on the static load data, the dynamic load data, and the variable load data to obtain load feature data;
[0012] The static load is the data value of all static loads :
[0013] ;
[0014] in, is the magnitude of the ith static load, n is the total number of loads;
[0015] The dynamic load data is the maximum value of the dynamic load. :
[0016] ;
[0017] in, is the dynamic load value at time t, T is the time interval;
[0018] The variable load data is the average value of the variable load within a certain time interval, that is, the time average load :
[0019] ;
[0020] in, is the variable load value at time t, T is the time interval;
[0021] The static load data, the dynamic load data and the variable load data are weighted and summed to obtain load characteristic data F:
[0022] ;
[0023] in, are weight values of the static load data, the dynamic load data and the variable load data respectively.
[0024] Furthermore, feature extraction is performed on the first monitoring data to obtain environmental feature data, which specifically includes the following steps:
[0025] Performing feature extraction on the first monitoring data to obtain temperature data, humidity data, wind speed data and atmospheric pressure data;
[0026] Normalizing the temperature data, the humidity data, the wind speed data, and the atmospheric pressure data to obtain normalized temperature data, normalized humidity data, normalized wind speed data, and normalized atmospheric pressure data;
[0027] The normalized temperature data, the normalized humidity data, the normalized wind speed data and the normalized atmospheric pressure data are subjected to data fusion processing to obtain a data fusion feature vector, and the data fusion feature vector is used as environmental feature data.
[0028] Furthermore, feature extraction is performed on the first monitoring data to obtain structural feature data, which specifically includes the following steps:
[0029] Performing feature extraction on the first monitoring data to obtain a strain data set;
[0030] Performing variance calculation on the strain data set to obtain a variance value of the strain data;
[0031] Performing mean calculation on the strain data set to obtain an average value of the strain data;
[0032] Extracting extreme values from the strain data set to obtain a maximum value of the strain data set and a minimum value of the strain data set;
[0033] The variance value of the strain data, the average value of the strain data, the maximum value of the strain data set and the minimum value of the strain data set are used to construct a feature vector to obtain structural feature data.
[0034] Furthermore, the polynomial regression prediction model is constructed using the following formula:
[0035] ;
[0036] Where Y is the displacement trend prediction result of the large-span spatial structure, X 1 、X 2 、X 3 They are environmental characteristic prediction data, load characteristic prediction data and structural characteristic prediction data. is the regression coefficient, is the error term.
[0037] Furthermore, after the step of obtaining the large-span spatial structure displacement trend prediction result, the method further includes:
[0038] Determine whether the displacement trend prediction result of the large-span spatial structure is within a preset displacement range;
[0039] When the displacement trend prediction result of the large-span space structure is within a preset range, a safety level assessment is performed on the large-span space structure based on the displacement trend prediction result of the large-span space structure;
[0040] When the displacement trend prediction result of the large-span spatial structure is not within a preset range, an early warning process is performed.
[0041] Furthermore, before the step of pre-building a vector autoregression prediction model, the method further includes:
[0042] Acquire a historical characteristic data set of a large-span spatial structure, wherein the historical characteristic data set includes historical environmental characteristic data, historical load characteristic data, and historical structural characteristic data;
[0043] Normalizing the historical feature data set of the large-span spatial structure to obtain a normalized historical feature data set of the large-span spatial structure;
[0044] The normalized historical feature data set of the large-span spatial structure is divided into a training feature data set and a test feature data set;
[0045] The vector autoregression prediction model is trained based on the training feature data set, and the vector autoregression prediction model is tested based on the test feature data set until the prediction accuracy of the vector autoregression prediction model meets a preset accuracy threshold or reaches a preset number of times, and then the testing of the vector autoregression prediction model is stopped.
[0046] Furthermore, the present invention also discloses a large-span spatial structure displacement trend prediction system, the system comprising:
[0047] A first monitoring data module is used to obtain monitoring data of the current large-span spatial structure and pre-process the monitoring data to obtain first monitoring data;
[0048] A feature extraction module, used to extract features from the first monitoring data to obtain environmental feature data, load feature data and structural feature data;
[0049] A feature prediction data module is used to pre-build a vector autoregression prediction model, and input the environmental feature data, the load feature data, and the structural feature data into the vector autoregression prediction model respectively, so that the vector autoregression prediction model outputs the environmental feature prediction data, the load feature prediction data, and the structural feature prediction data respectively;
[0050] The large-span spatial structure displacement trend prediction result module is used to construct a polynomial regression prediction model, input the environmental characteristic prediction data, the load characteristic prediction data and the structural characteristic prediction data into the polynomial regression prediction model, and obtain the large-span spatial structure displacement trend prediction result.
[0051] The beneficial effects of the present invention are:
[0052] The present invention can comprehensively consider the influence of environment, load and structural characteristics through the combination of vector autoregression prediction model and polynomial regression prediction model, and provide a more comprehensive displacement trend prediction. The vector autoregression prediction model is good at capturing the dynamic relationship between characteristics, and the polynomial regression model can handle complex nonlinear relationships, improve the accuracy of prediction and enhance reliability: multi-step prediction and comprehensive analysis help to identify potential structural problems, improve the accuracy of displacement prediction of large-span spatial structures, thereby improving the safety and reliability of large-span spatial structures. In addition, the method supports real-time processing and prediction of monitoring data, can detect anomalies in time, and enhance the real-time monitoring capability of large-span spatial structures. BRIEF DESCRIPTION OF THE DRAWINGS
[0053] Figure 1 A schematic diagram of a flow chart of a method for predicting displacement trends of a large-span spatial structure of the present application;
[0054] Figure 2 This is a schematic diagram of the structure of a large-span spatial structure displacement trend prediction system of the present application;
[0055] Figure 3 A schematic diagram of the structure of a computer device of the present application. DETAILED DESCRIPTION
[0056] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments.
[0057] Example 1, a method for predicting displacement trend of a large-span spatial structure, referring to Figure 1 The method includes steps S10 to S40. The detailed description of each step of the method is as follows:
[0058] S10. Acquire monitoring data of the large-span spatial structure, and pre-process the monitoring data to obtain first monitoring data.
[0059] In this embodiment, real-time or periodically recorded data are collected from the monitoring system of the large-span spatial structure, including but not limited to: environmental data, such as temperature, humidity, wind speed; various load data applied to the structure, such as vehicle load, equipment weight, etc.; structural data, such as actual response data of the structure, such as stress and strain. The acquired data undergoes a series of preprocessing steps to ensure that it is suitable for further analysis.
[0060] The preprocessing includes: data cleaning, i.e. identifying and processing errors or outliers in the data; data smoothing, applying smoothing techniques to reduce noise in the data and ensure data stability. The moving average method is used to smooth the data to eliminate short-term fluctuations caused by occasional changes or measurement errors; data standardization: converting the data into a unified format and range for consistent analysis. Standardization methods include scaling the data to a specific range (such as 0 to 1) so that data with different characteristics are on the same scale.
[0061] After the above processing steps, the cleaned and standardized data is the first monitoring data. These data have been adjusted to a format suitable for analysis and modeling, and can be effectively used for subsequent feature extraction, model training and predictive analysis.
[0062] S20. Perform feature extraction on the first monitoring data to obtain environmental feature data, load feature data and structural feature data.
[0063] In this embodiment, feature extraction is performed on the first monitoring data to obtain environmental feature data, which specifically includes the following steps:
[0064] Perform feature extraction on the first monitoring data to obtain temperature data, humidity data, wind speed data and atmospheric pressure data; normalize the temperature data, humidity data, wind speed data and atmospheric pressure data to obtain normalized temperature data, normalized humidity data, normalized wind speed data and normalized atmospheric pressure data; perform data fusion processing on the normalized temperature data, the normalized humidity data, the normalized wind speed data and the normalized atmospheric pressure data to obtain a data fusion feature vector, and use the data fusion feature vector as environmental feature data.
[0065] Extracting features from the first monitoring data to obtain load feature data specifically includes the following steps:
[0066] Feature extraction is performed on the first monitoring data to obtain static load data, dynamic load data and variable load data; weighted sum processing is performed on the static load data, the dynamic load data and the variable load data to obtain load feature data.
[0067] Performing feature extraction on the first monitoring data to obtain structural feature data specifically includes the following steps:
[0068] Perform feature extraction on the first monitoring data to obtain a strain data set; perform variance calculation on the strain data set to obtain a strain data variance value; perform mean calculation on the strain data set to obtain a strain data average value; extract extreme values from the strain data set to obtain a strain data set maximum value and a strain data set minimum value; construct a feature vector using the strain data variance value, the strain data average value, the strain data maximum value and the strain data set minimum value to obtain structural feature data.
[0069] S30, pre-constructing a vector autoregression prediction model, and inputting the environmental characteristic data, the load characteristic data, and the structural characteristic data into the vector autoregression prediction model respectively, so that the vector autoregression prediction model outputs the environmental characteristic prediction data, the load characteristic prediction data, and the structural characteristic prediction data respectively.
[0070] Specifically, in this embodiment, before the step of pre-building a vector autoregression prediction model, the following steps are also included:
[0071] Acquire a historical characteristic data set of a large-span spatial structure, wherein the historical characteristic data set includes historical environmental characteristic data, historical load characteristic data, and historical structural characteristic data;
[0072] Normalizing the historical feature data set of the large-span spatial structure to obtain a normalized historical feature data set of the large-span spatial structure;
[0073] The normalized historical feature data set of the large-span spatial structure is divided into a training feature data set and a test feature data set;
[0074] The vector autoregression prediction model is trained based on the training feature data set, and the vector autoregression prediction model is tested based on the test feature data set until the prediction accuracy of the vector autoregression prediction model meets a preset accuracy threshold or reaches a preset number of times, and then the testing of the vector autoregression prediction model is stopped.
[0075] The steps of pre-building the vector autoregression prediction model include: data preparation, model training and prediction result generation. By inputting environmental characteristic data, load characteristic data and structural characteristic data into the trained vector autoregression prediction model, the corresponding future prediction data is obtained. The dynamic relationship between variables is effectively captured to provide support for analysis and decision-making in related fields.
[0076] S40, constructing a polynomial regression prediction model, inputting the environmental characteristic prediction data, the load characteristic prediction data and the structural characteristic prediction data into the polynomial regression prediction model, and obtaining a displacement trend prediction result of a large-span spatial structure.
[0077] Specifically, in this embodiment, after the step of obtaining the large-span spatial structure displacement trend prediction result, the following steps are included:
[0078] Determine whether the displacement trend prediction result of the large-span spatial structure is within a preset displacement range;
[0079] When the displacement trend prediction result of the large-span space structure is within a preset range, a safety level assessment is performed on the large-span space structure based on the displacement trend prediction result of the large-span space structure;
[0080] When the displacement trend prediction result of the large-span spatial structure is not within a preset range, an early warning process is performed.
[0081] In this embodiment, feature extraction is performed on the first monitoring data to obtain environmental feature data, which specifically includes the following steps:
[0082] Performing feature extraction on the first monitoring data to obtain temperature data, humidity data, wind speed data and atmospheric pressure data;
[0083] Normalizing the temperature data, the humidity data, the wind speed data, and the atmospheric pressure data to obtain normalized temperature data, normalized humidity data, normalized wind speed data, and normalized atmospheric pressure data;
[0084] The normalized temperature data, the normalized humidity data, the normalized wind speed data and the normalized atmospheric pressure data are subjected to data fusion processing to obtain a data fusion feature vector, and the data fusion feature vector is used as environmental feature data.
[0085] Specifically, in this embodiment, the information related to temperature is extracted from the monitoring data by temperature records measured by sensors to obtain temperature values at different time points; similarly, records related to humidity are extracted based on humidity sensors; wind speed data is extracted based on wind speed sensors to record wind speeds at different time points; atmospheric pressure data is extracted to record atmospheric pressure levels at different time points.
[0086] The above data of different dimensions are converted to the same scale range for subsequent processing and fusion, so the temperature data, humidity data, wind speed data, and atmospheric pressure data are converted to the [0,1] interval or other standardized ranges. The temperature data, humidity data, wind speed data, and atmospheric pressure data are similarly processed to make them within a unified scale range; through this normalization process, the temperature, humidity, wind speed, and atmospheric pressure data can be converted to the same scale, which is helpful for subsequent comparison and fusion.
[0087] The normalized feature data are merged into a comprehensive feature vector to represent the overall environmental characteristics; the normalized feature data, namely the normalized temperature data, normalized humidity data, normalized wind speed data, and normalized atmospheric pressure data are combined into a vector, namely the data fusion feature vector: [normalized temperature, normalized humidity, normalized wind speed, normalized atmospheric pressure], each element represents the state of an environmental feature at the current time point, and such a feature vector can effectively describe the environmental conditions. The generated data fusion feature vector is further analyzed as environmental feature data, and these feature vectors are input into the prediction model, such as the vector autoregression prediction model, to predict and analyze the structural behavior.
[0088] In one embodiment, feature extraction is performed on the first monitoring data to obtain load feature data, which specifically includes the following steps: feature extraction is performed on the first monitoring data to obtain static load data, dynamic load data and variable load data; weighted summation is performed on the static load data, the dynamic load data and the variable load data to obtain load feature data.
[0089] In this embodiment, feature extraction is to extract useful information from the original data for analysis and modeling. In this process, the original monitoring data. That is, the first monitoring data is processed to extract load characteristic data. Static load data refers to a constant load that does not change over time during the monitoring period. Static loads usually come from long-term loads, such as the deadweight of a building or the weight of a resident object. In the above steps, feature extraction is performed on the first monitoring data to obtain load characteristic data, which specifically includes the following steps:
[0090] The static load is the data value of all static loads. :
[0091] ;
[0092] in, is the magnitude of the ith static load, and n is the total number of loads; dynamic load data refers to loads that change over time and are usually related to external factors such as wind, vibration, or traffic flow;
[0093] The dynamic load data is the maximum value of the dynamic load. :
[0094] ;
[0095] in, is the dynamic load value at time t, where T is the time interval; variable load data refers to loads that fluctuate greatly over time, including loads that change periodically or irregularly, but the amplitude of change is larger than that of static loads, such as crowd loads, temporary equipment loads, etc.;
[0096] The variable load data is the average value of the variable load within a certain time interval, that is, the time average load. :
[0097] ;
[0098] in, is the variable load value at time t, T is the time interval;
[0099] When the static load data, dynamic load data and variable load data are extracted, the next step is weighted summation processing, that is, the static load data, the dynamic load data and the variable load data are weighted summation processed to obtain the load characteristic data F:
[0100] ;
[0101] in, are weight values of the static load data, the dynamic load data and the variable load data respectively.
[0102] The process is to combine these load data into a unified load characteristic data. The specific steps are as follows: Assign different weights to each type of load data. These weights can be determined according to the needs of the actual application, based on professional knowledge, historical data analysis or other criteria. Add up the weighted load data to obtain a comprehensive load characteristic data. This step integrates all load information into a single data representing the overall load status.
[0103] Through the above feature extraction and weighted summation processing, the load characteristic data obtained is a comprehensive representation of the overall load state. This data is used for further analysis, modeling or decision support. The load data is divided into static, dynamic and variable loads through feature extraction, and then the data is weighted summed to generate a comprehensive load characteristic data. This processing method helps us convert complex load information into useful characteristic data to support further analysis and applications, such as structural health monitoring, load prediction, etc.
[0104] In this embodiment, feature extraction is performed on the first monitoring data to obtain structural feature data, which specifically includes the following steps:
[0105] Performing feature extraction on the first monitoring data to obtain a strain data set;
[0106] Performing variance calculation on the strain data set to obtain a variance value of the strain data;
[0107] Performing mean calculation on the strain data set to obtain an average value of the strain data;
[0108] Extracting extreme values from the strain data set to obtain a maximum value of the strain data set and a minimum value of the strain data set;
[0109] The variance value of the strain data, the average value of the strain data, the maximum value of the strain data set and the minimum value of the strain data set are used to construct a feature vector to obtain structural feature data.
[0110] In this embodiment, a strain data set is extracted from the first monitoring data. The strain data set contains the strain information of the structure at different time points and is an important indicator in structural health monitoring. By calculating the variance of the strain data set, the discreteness of the data is obtained, which reflects the volatility and stability of the strain value; by calculating the mean of the strain data set, the overall level of strain is represented; the maximum and minimum values in the strain data are extracted to reflect the extreme range of the strain data. The above calculation results, i.e., the variance value of the strain data, the mean value of the strain data, the maximum value and the minimum value of the strain data are combined into a feature vector. This feature vector integrates the main statistical characteristics of the strain data and represents the strain state of the structure.
[0111] Specifically, the strain data set is [0.12, 0.15, 0.10, 0.18, 0.14], the calculated variance is 0.00092, the calculated average value of the strain data is 0.138, the extreme values of the strain data are extracted, the maximum value is 0.18, and the minimum value is 0.10. The above calculation results are combined into a eigenvector, the eigenvector is [variance value, average value, maximum value, minimum value], that is, the eigenvector is [0.00092, 0.138, 0.18, 0.10], which integrates the key statistical characteristics of the strain data and can be used for further structural health analysis or modeling.
[0112] The structural feature data is a feature vector constructed through the above steps, which concentrates the key statistical information of the strain data in a vector. These features can be used for further analysis, such as structural health assessment, fault detection, or performance prediction. This process constructs a feature vector containing the main strain features by performing variance calculation, mean calculation, and extreme value extraction on the strain data. These structural feature data help describe and analyze the health status and performance of the structure.
[0113] In this embodiment, the polynomial regression prediction model is constructed using the following formula:
[0114] ;
[0115] Where Y is the displacement trend prediction result of the large-span spatial structure, X 1 、X 2、X 3 They are environmental characteristic prediction data, load characteristic prediction data and structural characteristic prediction data. is the regression coefficient, is the error term.
[0116] In this embodiment, the polynomial regression model captures the complex relationship of predicting the displacement trend of a large-span spatial structure by including linear terms, interaction terms, and quadratic terms of the independent variables. Specifically:
[0117] is a constant term, which represents the predicted value of Y when all independent variables X1, X2, and X3 are zero. The linear term Describes the direct effects of environmental characteristics, load characteristics, and structural characteristics on displacement trends; interaction terms Describes the effect of the interaction between different features on the displacement trend, the square term Describes the nonlinear effects of features; It is an error term, which represents random fluctuations that cannot be explained by the model. It captures the part of the real data that is not explained by the model. Through the combination of these terms, the model can fit the data more flexibly and capture complex nonlinear relationships, thereby providing more accurate predictions. Use historical data to train the polynomial regression model, estimate the regression coefficient β, input the environmental factors, load effects and structural deformation data obtained by real-time monitoring into the established polynomial regression model, generate real-time displacement prediction values, and provide early warning information based on the real-time updated displacement prediction results to support engineers in making timely structural maintenance, load adjustment or other decisions.
[0118] In this embodiment, after the step of obtaining the large-span spatial structure displacement trend prediction result, the following steps are also included:
[0119] Determine whether the displacement trend prediction result of the large-span spatial structure is within a preset displacement range;
[0120] When the displacement trend prediction result of the large-span spatial structure is within a preset range, a safety level assessment of the large-span spatial structure is performed based on the displacement trend prediction result of the large-span spatial structure; when the displacement trend prediction result of the large-span spatial structure is not within the preset range, early warning processing is performed.
[0121] In this embodiment, the preset range is set according to the design specification, historical data or engineering standard, and the preset range is ±1. When the displacement trend prediction result of the large-span spatial structure is within the preset range, a further safety level assessment is performed, including: analyzing the predicted displacement trend and its impact on the overall performance of the structure, and using the structural health assessment standard or model to determine the safety level of the structure. For example, the safety status of the structure is marked according to the level A, B, or C.
[0122] When the displacement trend prediction result of the large-span spatial structure is not within the preset range, early warning processing is performed, including issuing an alarm to notify relevant personnel of potential safety risks, conducting on-site inspections or detailed analysis to determine the actual cause of the problem, and taking repairs, reinforcements or other engineering measures based on the inspection results to ensure the safety of the structure.
[0123] In this embodiment, before the step of pre-building a vector autoregression prediction model, the following steps are also included:
[0124] Acquire a historical characteristic data set of a large-span spatial structure, wherein the historical characteristic data set includes historical environmental characteristic data, historical load characteristic data, and historical structural characteristic data;
[0125] Normalizing the historical feature data set of the large-span spatial structure to obtain a normalized historical feature data set of the large-span spatial structure;
[0126] The normalized historical feature data set of the large-span spatial structure is divided into a training feature data set and a test feature data set;
[0127] The vector autoregression prediction model is trained based on the training feature data set, and the vector autoregression prediction model is tested based on the test feature data set until the prediction accuracy of the vector autoregression prediction model meets a preset accuracy threshold or reaches a preset number of times, and then the testing of the vector autoregression prediction model is stopped.
[0128] In this embodiment, historical data is collected for model training and testing. The historical environmental characteristic data includes environmental conditions such as temperature data, humidity data, wind speed data, and atmospheric pressure data; the historical load characteristic data includes static load data, dynamic load data, and variable load data load history records; the historical structural characteristic data includes strain data sets.
[0129] Normalize the historical feature data set, prepare training and test data to evaluate the performance of the model, divide the normalized historical feature data set into a training feature data set and a test feature data set, the training feature data set is used to train the model, and the test feature data set is used to test the prediction ability of the model, build the model through the training data set, verify the accuracy of the model through the test data set, and use the training feature data set to train the vector autoregression prediction model.
[0130] The vector autoregression prediction model is a multivariate time series prediction model that can capture the dynamic relationship between different features. The model is tested using a test feature data set to evaluate its prediction accuracy, which can be evaluated by calculating the error between the predicted value and the actual value, defining a preset accuracy standard. In this embodiment, the preset condition is that the mean square error is less than the threshold, and the maximum number of training times or rounds is set to avoid overfitting and ensure the training stability of the model. During the model training process, its performance on the test data set is monitored, and training is stopped when the model's prediction accuracy meets the preset threshold or reaches the preset number of training times. Specifically, if the mean square error on the test data set is less than the preset 0.01, it is considered that the model accuracy has met the requirements.
[0131] These steps ensure that an effective vector autoregression prediction model is constructed from historical data, and through normalization, data partitioning and accuracy testing, the model can accurately predict the displacement trend of large-span spatial structures. The specific steps include acquiring and processing data, training and testing the model until the accuracy requirements are met or the preset number of times is reached.
[0132] Example 2, a large-span spatial structure displacement trend prediction system, referring to Figure 2 The system applies a method for predicting displacement trend of a large-span spatial structure described in Example 1, and the system comprises:
[0133] The first monitoring data module 10 is used to obtain monitoring data of the current large-span spatial structure and pre-process the monitoring data to obtain first monitoring data;
[0134] A feature extraction module 20 is used to extract features from the first monitoring data to obtain environmental feature data, load feature data and structural feature data;
[0135] The feature prediction data module 30 is used to pre-build a vector autoregression prediction model, and input the environmental feature data, the load feature data and the structural feature data into the vector autoregression prediction model respectively, so that the vector autoregression prediction model outputs the environmental feature prediction data, the load feature prediction data and the structural feature prediction data respectively;
[0136] The large-span spatial structure displacement trend prediction result module 40 is used to construct a polynomial regression prediction model, input the environmental characteristic prediction data, the load characteristic prediction data and the structural characteristic prediction data into the polynomial regression prediction model, and obtain the large-span spatial structure displacement trend prediction result.
[0137] It can be understood that the various components of the large-span spatial structure displacement trend prediction system proposed in the present application can realize the functions of any of the large-span spatial structure displacement trend prediction methods described above.
[0138] In this embodiment, the feature extraction module 20 further includes executing the following steps:
[0139] Performing feature extraction on the first monitoring data to obtain temperature data, humidity data, wind speed data and atmospheric pressure data;
[0140] Normalizing the temperature data, the humidity data, the wind speed data, and the atmospheric pressure data to obtain normalized temperature data, normalized humidity data, normalized wind speed data, and normalized atmospheric pressure data;
[0141] The normalized temperature data, the normalized humidity data, the normalized wind speed data and the normalized atmospheric pressure data are subjected to data fusion processing to obtain a data fusion feature vector, and the data fusion feature vector is used as environmental feature data.
[0142] In this embodiment, the feature extraction module 20 further includes executing the following steps:
[0143] Extracting features from the first monitoring data to obtain static load data, dynamic load data and variable load data;
[0144] The static load data, the dynamic load data and the variable load data are weighted and summed to obtain load characteristic data.
[0145] In this embodiment, the feature extraction module 20 further includes executing the following steps:
[0146] Performing feature extraction on the first monitoring data to obtain a strain data set;
[0147] Performing variance calculation on the strain data set to obtain a variance value of the strain data;
[0148] Performing mean calculation on the strain data set to obtain an average value of the strain data;
[0149] Extracting extreme values from the strain data set to obtain a maximum value of the strain data set and a minimum value of the strain data set;
[0150] The variance value of the strain data, the average value of the strain data, the maximum value of the strain data set and the minimum value of the strain data set are used to construct a feature vector to obtain structural feature data.
[0151] In this embodiment, the feature prediction data module 30 pre-builds a vector autoregression prediction model based on the following formula:
[0152] ;
[0153] Where Y is the displacement trend prediction result of the large-span spatial structure,X 1 、X 2 、X 3 They are environmental characteristic prediction data, load characteristic prediction data and structural characteristic prediction data. is the regression coefficient, is the error term.
[0154] In this embodiment, the large-span spatial structure displacement trend prediction result module 40 further includes executing the following steps:
[0155] Determine whether the displacement trend prediction result of the large-span spatial structure is within a preset displacement range;
[0156] When the displacement trend prediction result of the large-span space structure is within a preset range, a safety level assessment is performed on the large-span space structure based on the displacement trend prediction result of the large-span space structure;
[0157] When the displacement trend prediction result of the large-span spatial structure is not within a preset range, an early warning process is performed.
[0158] In one embodiment, the large-span spatial structure displacement trend prediction result module 40 further includes executing the following steps:
[0159] Acquire a historical characteristic data set of a large-span spatial structure, wherein the historical characteristic data set includes historical environmental characteristic data, historical load characteristic data, and historical structural characteristic data;
[0160] Normalizing the historical feature data set of the large-span spatial structure to obtain a normalized historical feature data set of the large-span spatial structure;
[0161] The normalized historical feature data set of the large-span spatial structure is divided into a training feature data set and a test feature data set;
[0162] The vector autoregression prediction model is trained based on the training feature data set, and the vector autoregression prediction model is tested based on the test feature data set until the prediction accuracy of the vector autoregression prediction model meets a preset accuracy threshold or reaches a preset number of times, and then the testing of the vector autoregression prediction model is stopped.
[0163] In this embodiment, the feature extraction module further includes executing the following steps:
[0164] Performing feature extraction on the first monitoring data to obtain temperature data, humidity data, wind speed data and atmospheric pressure data;
[0165] Normalizing the temperature data, the humidity data, the wind speed data, and the atmospheric pressure data to obtain normalized temperature data, normalized humidity data, normalized wind speed data, and normalized atmospheric pressure data;
[0166] The normalized temperature data, the normalized humidity data, the normalized wind speed data and the normalized atmospheric pressure data are subjected to data fusion processing to obtain a data fusion feature vector, and the data fusion feature vector is used as environmental feature data.
[0167] In this embodiment, the feature extraction module further includes executing the following steps:
[0168] Used to extract features from the first monitoring data to obtain static load data, dynamic load data and variable load data;
[0169] It is used to perform weighted sum processing on the static load data, the dynamic load data and the variable load data to obtain load characteristic data.
[0170] Embodiment 3, a computer device, referring to Figure 3 , the system is implanted into the computer device, and the internal structure of the computer device can be as follows Figure 3 As shown. The computer device includes a processor, a memory, a network interface, a display device and an input device connected through a system bus. Among them, the network interface of the computer device is used to communicate with an external terminal through a network connection. The display device of the computer device is used to display an interactive page. The input device of the computer device is used to receive user input. The processor designed for the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium. The non-volatile storage medium stores an operating system, a computer program and a database. The database of the computer device is used to store raw data. When the computer program is executed by the processor, a large-span spatial structure displacement trend prediction method described in Example 1 is implemented.
[0171] Embodiment 4, a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by the processor, implements the method for predicting displacement trends of large-span spatial structures described in Embodiment 1.
[0172] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory.
[0173] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.
Claims
1. A method for predicting displacement trend of a large-span spatial structure, characterized in that: The steps include: Acquire monitoring data of the large-span spatial structure, and pre-process the monitoring data to obtain first monitoring data; Extracting features from the first monitoring data to obtain environmental feature data, load feature data, and structural feature data; Pre-constructing a vector autoregression prediction model, and inputting the environmental characteristic data, the load characteristic data, and the structural characteristic data into the vector autoregression prediction model respectively, so that the vector autoregression prediction model outputs the environmental characteristic prediction data, the load characteristic prediction data, and the structural characteristic prediction data respectively; Constructing a polynomial regression prediction model, inputting the environmental characteristic prediction data, the load characteristic prediction data and the structural characteristic prediction data into the polynomial regression prediction model, and obtaining a large-span spatial structure displacement trend prediction result; In the above steps, feature extraction is performed on the first monitoring data to obtain load feature data, which specifically includes the following steps: Extracting features from the first monitoring data to obtain static load data, dynamic load data, and variable load data; performing weighted sum processing on the static load data, the dynamic load data, and the variable load data to obtain load feature data; The static load is the data value of all static loads : ; in, is the magnitude of the ith static load, n is the total number of loads; The dynamic load data is the maximum value of the dynamic load. : ; in, is the dynamic load value at time t, T is the time interval; The variable load data is the average value of the variable load within a certain time interval, that is, the time average load : ; in, is the variable load value at time t, T is the time interval; The static load data, the dynamic load data and the variable load data are weighted and summed to obtain load characteristic data F: ; in, are weight values of the static load data, the dynamic load data and the variable load data respectively.
2. A method for predicting displacement trend of a large-span spatial structure according to claim 1, characterized in that: Extracting features from the first monitoring data to obtain environmental feature data specifically includes the following steps: Performing feature extraction on the first monitoring data to obtain temperature data, humidity data, wind speed data and atmospheric pressure data; Normalizing the temperature data, the humidity data, the wind speed data, and the atmospheric pressure data to obtain normalized temperature data, normalized humidity data, normalized wind speed data, and normalized atmospheric pressure data; The normalized temperature data, the normalized humidity data, the normalized wind speed data and the normalized atmospheric pressure data are subjected to data fusion processing to obtain a data fusion feature vector, and the data fusion feature vector is used as environmental feature data.
3. A method for predicting displacement trend of a large-span spatial structure according to claim 1, characterized in that: Performing feature extraction on the first monitoring data to obtain structural feature data specifically includes the following steps: Performing feature extraction on the first monitoring data to obtain a strain data set; Performing variance calculation on the strain data set to obtain a variance value of the strain data; Performing mean calculation on the strain data set to obtain an average value of the strain data; Extracting extreme values from the strain data set to obtain a maximum value of the strain data set and a minimum value of the strain data set; The variance value of the strain data, the average value of the strain data, the maximum value of the strain data set and the minimum value of the strain data set are used to construct a feature vector to obtain structural feature data.
4. A method for predicting displacement trend of a large-span spatial structure according to claim 1, characterized in that: The polynomial regression prediction model is constructed using the following formula: ; Where Y is the displacement trend prediction result of the large-span spatial structure, X 1 、X 2 、X 3 They are environmental characteristic prediction data, load characteristic prediction data and structural characteristic prediction data. is the regression coefficient, is the error term.
5. The method for predicting displacement trend of a large-span spatial structure according to claim 1, characterized in that: After the step of obtaining the large-span spatial structure displacement trend prediction result, the method further includes: Determine whether the displacement trend prediction result of the large-span spatial structure is within a preset displacement range; When the displacement trend prediction result of the large-span space structure is within a preset range, a safety level assessment is performed on the large-span space structure based on the displacement trend prediction result of the large-span space structure; When the displacement trend prediction result of the large-span spatial structure is not within a preset range, an early warning process is performed.
6. A method for predicting displacement trend of a large-span spatial structure according to claim 1, characterized in that: Before the step of pre-building a vector autoregression prediction model, the method further includes: Acquire a historical characteristic data set of a large-span spatial structure, wherein the historical characteristic data set includes historical environmental characteristic data, historical load characteristic data, and historical structural characteristic data; Normalizing the historical feature data set of the large-span spatial structure to obtain a normalized historical feature data set of the large-span spatial structure; The normalized historical feature data set of the large-span spatial structure is divided into a training feature data set and a test feature data set; The vector autoregression prediction model is trained based on the training feature data set, and the vector autoregression prediction model is tested based on the test feature data set until the prediction accuracy of the vector autoregression prediction model meets a preset accuracy threshold or reaches a preset number of times, and then the testing of the vector autoregression prediction model is stopped.
7. A large-span spatial structure displacement trend prediction system, characterized in that: The system applies a method for predicting displacement trend of a large-span spatial structure as claimed in any one of claims 1 to 6, and the system comprises: A first monitoring data module is used to obtain monitoring data of the current large-span spatial structure and pre-process the monitoring data to obtain first monitoring data; A feature extraction module, used to extract features from the first monitoring data to obtain environmental feature data, load feature data and structural feature data; A feature prediction data module is used to pre-build a vector autoregression prediction model, and input the environmental feature data, the load feature data, and the structural feature data into the vector autoregression prediction model respectively, so that the vector autoregression prediction model outputs the environmental feature prediction data, the load feature prediction data, and the structural feature prediction data respectively; The large-span spatial structure displacement trend prediction result module is used to construct a polynomial regression prediction model, input the environmental characteristic prediction data, the load characteristic prediction data and the structural characteristic prediction data into the polynomial regression prediction model, and obtain the large-span spatial structure displacement trend prediction result.
