Method for predicting displacement response of tower crane under typhoon action by fusing data and model

Abstract

The invention relates to a data and model fused tower crane displacement response prediction method under the typhoon action, which comprises the following steps: constructing a monitoring system, acquiring actual measurement data of a construction site and a tower crane, and constructing a tower crane finite element model by utilizing a design parameter set of the tower crane; the extreme wind speed under the most unfavorable wind direction of a construction site is obtained based on typhoon full-path simulation and wind field simulation, and the tower crane finite element model is calibrated by analyzing a parameter object needing to be calibrated in the tower crane finite element model through sensitivity, so that the calibrated tower crane finite element model can accurately reflect the response condition of a tower crane structure; a calibrated tower crane finite element model is used for simulating a data set formed by displacement data under the typhoon working condition to serve as a training set, the defect of an extreme weather sample in an actually-measured training sample is overcome, then response prediction of tower crane displacement under the typhoon working condition is achieved by training a CNN-BiLSTM-Adaboost displacement response prediction model, and the method is simple and convenient to implement. The precision and robustness of tower crane displacement prediction under the extreme typhoon working condition are remarkably improved, and sufficient emergency processing time is reserved for a construction site.

Description

Technical Field

[0001] The present invention relates to the technical field of tower crane displacement safety monitoring, and in particular to a method for predicting tower crane displacement response under typhoon by integrating data and models. Background Art

[0002] With the increase in high-rise buildings, tower cranes have become indispensable heavy equipment on construction sites. Due to their high and flexible structural characteristics, their stability will face great challenges in the face of extreme weather, especially the strong winds and rainstorms brought by typhoons. In my country's coastal areas, where typhoons frequently occur, tower crane collapse accidents occur from time to time, usually causing large casualties and economic losses; the strong winds and rainstorms brought by typhoons undoubtedly pose a great threat to the stability of tower cranes in coastal areas. Therefore, carrying out safety monitoring of tower cranes and early warning and prediction of dynamic response under the action of extreme typhoons is of great significance to improving construction safety and reducing the accident rate.

[0003] In the field of tower crane safety monitoring, many scholars have developed advanced IoT sensing monitoring technologies. Among them, Zhong et al. proposed a tower crane group safety management system (SMS-TC) that combines wireless sensor networks and IoT technologies to improve the safety and efficiency of tower crane operations at construction sites and reduce construction risks through real-time monitoring and anti-collision algorithms; Wu et al. developed an IoT platform that can monitor crane groups in real time and identify potential safety risks. In recent years, the introduction of image recognition technology has further enriched monitoring methods and improved the accuracy and response speed of monitoring. For example, Xu analyzed the lifting operation safety behavior of workers at prefabricated construction sites using computer vision technology and proposed an effective method for identifying unsafe behaviors; Yang et al. proposed a new safety distance monitoring method, which collects video data through tower crane cameras and uses Mask R-CNN for image recognition to evaluate unsafe behaviors in monitoring; Wang et al. developed a real-time positioning system based on computer vision technology to achieve accurate three-dimensional positioning of tower cranes and hooks. In addition, some innovative technologies have also been initially applied on construction sites. For example, Zhang and Pan proposed virtual reality (VR) technology, which enables workers to immersively experience various safety risks that may be encountered during the construction process; He et al. proposed augmented reality (AR) technology that can superimpose real-time data or instructions on the crane's control device to help operators understand the data and interact with the equipment, so as to make wise and accurate decisions when performing maintenance and production tasks. However, VR and AR systems are costly in hardware, software development and operator training, and may not be applicable to small-scale engineering units with limited resources. Overall, although significant achievements have been made in real-time monitoring of tower cranes, there are still some shortcomings in current research and engineering applications in terms of prediction and early warning. The current monitoring system lacks reliable early warning capabilities and cannot accurately predict the structural response trend. When the structure faces potential risks, it is impossible to reserve sufficient emergency response time for the construction site, which increases the risk of construction safety. Therefore, it is necessary to carry out tower crane response prediction and early warning work. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to accurately predict the displacement response data of the tower crane under typhoon weather and reserve sufficient emergency processing time for the construction site.

[0005] The present invention provides a method for predicting the displacement response of a tower crane under the action of a typhoon by fusing data and a model. The tower crane comprises a tower body, a balance arm, a lifting arm, a tie rod connecting the tower top and the balance arm, and connecting the tower top and the lifting arm. The wind speed environment faced by the tower crane comprises a typhoon condition and a benign wind condition. The method for predicting the displacement response of a tower crane under the action of a typhoon by fusing data and a model comprises: Step 1. Construct a monitoring system for real-time monitoring of the wind speed and displacement data of the tower crane at the construction site, wherein the monitoring system includes a plurality of inclination sensors arranged along the height direction of the tower body, an acceleration sensor arranged at the junction of the tower body and the crane arm, and an anemometer arranged at the construction site, wherein the inclination sensor is used to collect inclination data to calculate the displacement data of the tower crane, and the anemometer is used to collect the wind speed at the construction site; modal analysis is performed on the acceleration data collected by the acceleration sensor to obtain the modal frequency and vibration type of the tower crane; Step 2. Obtain the design parameter set of the tower crane, and construct a finite element model of the tower crane based on the design parameter set; divide the finite element model of the tower crane into N simulation areas on average, and determine the simulation nodes of each simulation area; and calculate the extreme wind speed under the most unfavorable wind direction of the construction site as the upper limit of the simulated wind speed based on the typhoon full path simulation and wind field simulation; randomly sample R wind speed data in the range of 0 to extreme wind speed to form a wind speed sequence for a continuous period to form a simulated wind speed set; Step 3. Perform sensitivity analysis on the parameter objects in the design parameter set, and determine the parameter objects that need to be calibrated in the tower crane finite element model according to the sensitivity analysis results; Step 4. Calibrate the tower crane finite element model based on the parameter object that needs to be calibrated, and perform modal analysis on the calibrated finite element model to obtain the modal frequency and vibration mode of the calibrated tower crane finite element model; determine whether the vibration mode of the tower crane finite element model and the tower crane in the same mode is consistent, and whether the difference between the modal frequency of the tower crane and the modal frequency of the tower crane finite element model is less than a preset threshold value K. If so, proceed to step 5; if not, return to step 4; Step 5. Calculate the wind load time history of each simulation node based on the simulated wind speed, apply the wind load time history of each simulation node to the calibrated tower crane finite element model along the most unfavorable direction to obtain displacement data, and form wind speed-displacement data. Thus, based on the simulated wind speed set, obtain R groups of wind speed-displacement data sets containing continuous time periods; Step 6. Perform exponential fitting on the data set obtained in step 5 to obtain a simulated data fitting line; obtain the wind speed and tower crane displacement of the construction site in the non-working state before the typhoon from the monitoring system, and perform exponential fitting to obtain the measured data fitting line; determine whether the difference between the measured data fitting line and the simulated data fitting line in the tower crane displacement at the same wind speed is less than a preset threshold value O. If so, the calibrated tower crane finite element model matches the tower crane at the construction site and proceeds to step 7; if not, return to step 4; Step 7. Construct a CNN-BiLSTM-Adaboost displacement response prediction model; Step 8. Use the data set obtained in step 5 as the training set, the wind speed-displacement of the current period as the input, and the displacement of the next period as the output to train the CNN-BiLSTM-Adaboost displacement response prediction model; obtain V groups of wind speed-displacement composition test sets of continuous periods under typhoon conditions from the monitoring system, and test the trained CNN-BiLSTM-Adaboost displacement response prediction model based on the test set to realize the response prediction of tower crane displacement under typhoon conditions.

[0006] Compared with the prior art, the present application has the following advantages: the present application first constructs a monitoring system for monitoring the wind speed and tower crane displacement data at the construction site, obtains the measured data of the construction site and the tower crane, and then uses the design parameter set of the tower crane to construct a finite element model of the tower crane. At the same time, the typhoon full-path simulation and wind field simulation are used to obtain the extreme wind speed under the most unfavorable wind direction at the construction site. The parameter objects that need to be calibrated are determined by performing sensitivity analysis on the parameter objects in the design parameter set, and the tower crane finite element model is calibrated based on the parameter objects that need to be calibrated, so that the calibrated tower crane finite element model can accurately reflect the response of the tower crane structure; then, the calibrated tower crane finite element model is used to simulate the displacement data composed of the displacement data under typhoon conditions as a training set to make up for the lack of extreme weather samples in the measured training samples, and then the CNN-BiLSTM-Adaboost displacement response prediction model is trained to realize the response prediction of the tower crane displacement under typhoon conditions, which significantly improves the accuracy and robustness of the tower crane displacement prediction under extreme typhoon conditions, and reserves sufficient emergency processing time for the construction site.

[0007] In one possible implementation, the monitoring system constructed in step 1 includes a perception layer, a network layer and an application layer that are communicatively connected in sequence, the perception layer includes an inclinometer, an anemometer and an acceleration sensor, the anemometer is used to collect wind direction data and wind speed data at the construction site in real time; the network layer includes a GPRS network for realizing wireless transmission of data, and the application layer is composed of a remote monitoring cloud platform for displaying real-time data collected by the perception layer.

[0008] In a possible implementation, the multiple inclination sensors arranged along the height direction of the tower body in step 1 divide the tower body into multiple monitoring segments, and each of the inclination sensors collects the three-axis inclination of the corresponding monitoring segment on the X-axis, Y-axis, and Z-axis.

[0009] In a possible implementation, the method for calculating the displacement data of the tower crane at the construction site by the monitoring system includes: First, the horizontal plane inclination angles of the inclination sensor along the X-axis and the Y-axis are obtained in turn; Next, combined with the height position of each inclination sensor on the tower body, the coordinates of each inclination sensor are calculated through the conversion formula, thereby obtaining the horizontal offset of the tower body position where each inclination sensor is located; the conversion formula is: ; In the formula, is the height position of each inclination sensor, are the horizontal plane inclination angles of the X-axis and Y-axis of each inclination sensor respectively; Finally, the horizontal offset of each tilt sensor is added in turn to obtain the total horizontal offset of the tower top as the displacement data of the tower crane.

[0010] In a possible implementation manner, the calculation formula for performing sensitivity analysis on the design parameter set in step 3 is: ; In the formula, represents the parameter sensitivity, It is the parameter object in the tower crane design parameters. is the parameter value change of the parameter object in the tower crane design parameter set, is the output of the tower crane finite element model, It is the output change of the parameter object in the tower crane finite element model before and after the change; Based on the sensitivity analysis results, the parameter with the largest sensitivity value in the design parameter set of the tower crane is determined as the parameter object that needs to be calibrated, and the corresponding parameters in the finite element model of the tower crane are calibrated so that the modal frequency of the finite element model of the tower crane under the same vibration mode matches the modal frequency of the tower crane at the construction site.

[0011] In a possible implementation, the CNN-BiLSTM-Adaboost structural response prediction model constructed in step 7 integrates a CNN model, a BiLSTM model and an Adaboost algorithm.

[0012] In a possible implementation, the CNN model is used to extract image features from wind speed-displacement, and output the extracted image features to the BiLSTM model. The CNN model includes an input layer, a convolution layer, a pooling layer, a fully connected layer, and an output layer connected in sequence. The convolution layer is used to capture local features of the input data and form a feature map. The convolution layer is expressed as: ; In the formula, represents the convolutional layer Output feature map, represents a nonlinear activation function, is the weight matrix, Input features to the convolutional layer, represents the deviation vector; The pooling layer is used to perform downsampling operations to reduce the dimension of the data; the pooling layer is expressed as: ; In the formula, represents the output of the pooling layer, Represents the bias of the pooling layer; The fully connected layer is used to integrate and transform the features extracted by the convolution layer and the pooling layer, and perform the final classification and regression tasks to output the prediction results; the fully connected layer is expressed as: ; In the formula, Represents the final output result, represents the weight matrix, represents the bias of the fully connected layer.

[0013] In a possible implementation, the BiLSTM model is used to perform sequence modeling on the image features. The BiLSTM model includes an input layer, a forward LSTM model, a reverse LSTM model, and an output layer. The data input from the input layer to the BiLSTM model is captured through the two forward LSTM models and the reverse LSTM model, and the sequence dynamics of the image features are output from the output layer. The BiLSTM model is expressed as: ; ; ; In the formula, Indicates that the forward LSTM model is in 1~ n +1 hidden layer state at time; Indicates that the reverse LSTM model is in 1~ n +1 hidden layer state at time; Represents the inter-layer activation function; Indicates 1~ n Input data at time, represents the weight between the input layer and the forward LSTM model, represents the weight between the input layer and the reverse LSTM model, represents the weight of the forward LSTM model, represents the weight of the reverse LSTM model, represents the weights of the reverse LSTM model and the output layer, Represents the weights of the forward LSTM model and the output layer; Indicates 1~ n Output data at the moment.

[0014] In a possible implementation, the Adaboost algorithm constructs a strong classifier by integrating multiple weak classifiers, and outputs a prediction result as the displacement data of the tower crane under typhoon conditions in the future period by iteratively adjusting the weights of the weak classifiers; the Adaboost algorithm is expressed as: ; ; ; ; In the formula, express Group training set data, is expressed as the normalization factor, represents the input vector and the output vector, represents a weak classifier, express The weight of represents the loss error, represents the sample weight, Represents a strong classifier. In each iteration, the weak classifier Weight According to its loss error To calculate, then, the weak classifier Performance on training set data, sample weights Update, and finally, the weak classifier Ensemble into a strong classifier middle.

[0015] In a possible implementation, the calculation of the wind load time series of each simulation node based on the simulated wind speed in step 5 specifically includes: firstly calculating the wind speed time series of each simulation node using the harmonic superposition method for the simulated wind speed; and then calculating the wind load time series of each simulation node using the wind load formula based on the wind speed time series of each simulation node. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 A schematic diagram of a self-elevating tower crane QTZ63 used in a specific embodiment of the present invention; Figure 3 is a system structure diagram of the monitoring system of the present invention; Figure 4 It is a layout diagram of the tilt sensor of a specific embodiment of the present invention; Figure 5 It is a schematic diagram of the division of simulation areas and simulation nodes on a tower crane in a specific embodiment of the present invention; Figure 6 The first-order vibration mode diagram of the finite element model of the tower crane according to the specific embodiment of the present invention, part (a) is a schematic diagram of the first-order vibration mode of the tower crane; part (b) is a schematic diagram of the first vibration mode of the boom; Figure 7 The second-order vibration mode diagram of the finite element model of the tower crane according to the specific embodiment of the present invention, part (a) is a schematic diagram of the second-order vibration mode of the tower crane; part (b) is a schematic diagram of the second-order vibration mode of the boom; Figure 8 The third-order vibration mode diagram of the finite element model of the tower crane according to the specific embodiment of the present invention, part (a) is a schematic diagram of the third-order vibration mode of the tower crane; part (b) is a schematic diagram of the second-order vibration mode of the boom; Fig. 9 This is a schematic diagram of the measured data fitting line and the simulated data fitting line obtained by fitting in step 6 of a specific embodiment of the present invention; Fig.10 This is a wind climate assessment result for a certain region according to a specific embodiment of the present invention, wherein part (a) represents the probability distribution of the annual maximum wind speed of each wind direction; part (b) represents the extreme wind speed distribution diagram of each wind direction under a 100-year return period; Fig.11 Part (a) is a comparison chart of the predicted values ​​and the measured values ​​of the training set (RMSE=9.7613); part (b) is a comparison chart of the predicted values ​​and the measured values ​​of the test set (RMSE=11.7857); part (c) is a regression chart of the training set; part (d) is a regression chart of the test set; and part (e) is a prediction error chart. DETAILED DESCRIPTION

[0017] First, those skilled in the art should understand that these implementations are only used to explain the technical principles of the embodiments of the present application, and are not intended to limit the protection scope of the embodiments of the present application. Those skilled in the art can make adjustments to them as needed to adapt to specific application scenarios.

[0018] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0019] In the embodiments of the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0020] The present application is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] See also Figure 2 As shown, the tower crane involved in the embodiment of the present application includes a tower body, a balance arm, a lifting arm, a tie rod connecting the tower top and the balance arm, and connecting the tower top and the lifting arm. The bottom of the tower body has a large rigidity, so it is treated as a fixed support; the root of the lifting arm, the connection between the lifting arm and the two tie rods, the connection between the balance arm and the two tie rods, and the connection between the tie rod and the tower top are all fixed hinge supports; the connection between the slewing support and the tower top and the tower body is fixed. The tower crane of this specific embodiment takes a self-elevating tower crane QTZ63 at a construction site in a certain area as the research object, such as Figure 2 As shown, the tower body is 51m high, using an integral standard section of 1.6m×1.6m×2.5m, the boom length is 50m, and the lifting moment is 630kN·m.

[0022] See also Figure 1 The specific embodiment of the present application discloses a method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model. In this specific embodiment, a wind speed greater than 20 m / s is defined as a typhoon condition, and a wind speed less than or equal to 20 m / s is defined as a benign wind condition, specifically including: Step 1. Build a monitoring system for real-time monitoring of wind speed and crane displacement data at the construction site, such as Figure 3 As shown, the monitoring system includes a perception layer, a network layer and an application layer connected in sequence through communication. The perception layer includes an inclination sensor for collecting the three-axis inclination of the tower body on the X-axis, Y-axis and Z-axis, an acceleration sensor for collecting acceleration data and an anemometer for collecting wind direction data and wind speed data at the construction site; the network layer includes a GPRS network for realizing wireless transmission of data, and the application layer is composed of a remote monitoring cloud platform for displaying real-time data collected by the perception layer, including displacement data of the tower crane at the construction site and wind speed and direction at the construction site; the acceleration sensor is arranged at the junction of the tower body and the crane arm. The distribution of the inclination sensor on the tower body in this specific embodiment is as shown in FIG. Figure 4The tilt sensors numbered 1 to 6 are sequentially arranged at different heights of the tower body, and the tilt sensor numbered 7 is arranged at the top of the tower; the multiple tilt sensors arranged along the height direction of the tower body divide the tower body into multiple monitoring segments, and each of the tilt sensors collects the three-axis tilt angles of the corresponding monitoring segment on the X-axis, Y-axis, and Z-axis; This specific embodiment uses an anemometer and an inclination sensor to study the correlation between wind speed and tower crane displacement. The sampling frequency of the anemometer is 4 Hz, and the changes in wind speed and wind direction of the site are automatically recorded in real time, and then the data is sent to the application layer through the gateway. The transmission interval is 1 minute. This specific embodiment collects the average wind speed of 10 minutes and the maximum 3s gust wind speed within 10 minutes; After the application layer obtains the tilt sensor data, the calculation method for calculating the displacement data of the tower crane includes: First, the horizontal plane inclination angles of the inclination sensor along the X-axis and the Y-axis are obtained in turn; Next, combined with the height position of each inclination sensor on the tower body, the coordinates of each inclination sensor are calculated through the conversion formula, thereby obtaining the horizontal offset of the tower body position where each inclination sensor is located; the conversion formula is: ; In the formula, is the height position of each inclination sensor, are the horizontal plane inclination angles of the X-axis and Y-axis of each inclination sensor respectively; Finally, the horizontal offset of each tilt sensor is added in turn to obtain the total horizontal offset of the tower top as the displacement data of the tower crane.

[0023] Step 2. Obtain the design parameter set of the tower crane, construct a finite element model of the tower crane in the spatial finite element analysis software based on the design parameter set, divide the finite element model of the tower crane into N simulation areas on average, and determine the simulation nodes of each simulation area; in this specific embodiment, the tower crane is divided into 18 sections, and the bolt nodes are determined as simulation nodes, such as Figure 5 As shown; and based on the typhoon full path simulation and wind field simulation, the extreme wind speed under the most unfavorable wind direction of the construction site is calculated as the upper limit of the simulated wind speed; R wind speed data are randomly sampled in the range of 0 to the extreme wind speed to form a wind speed sequence for a continuous period to form a simulated wind speed set; this specific embodiment constructs a total of R=1035 sets of data sets; This specific embodiment is based on the typhoon full-path simulation and wind field simulation method, generates 600-year near-ground typhoon wind speed samples that affect the area where the construction site is located, classifies them according to wind direction, and uses a multidimensional t-Copula function to construct a joint probability distribution function of extreme wind speeds in multiple wind directions, determines the extreme wind speed in each wind direction, and takes the extreme wind speed in the most unfavorable wind direction as the upper limit of the simulated wind speed, that is, the extreme wind speed. The typhoon full-path simulation and wind field simulation to calculate the extreme wind speed at the construction site are existing technologies and will not be described in detail here. In order to ensure sufficient annual extreme values ​​for statistical analysis, this specific embodiment conducted a 600-year Monte Carlo simulation to estimate the typhoon climate in a certain area. Using the simulated typhoon wind speed and the constant wind speed recorded by the weather station, based on the t-Copula function, a joint probability distribution model of the normal wind and the typhoon annual maximum wind speed series under 16 wind directions in the corresponding area of ​​the construction site was constructed; and the probability distribution of the annual maximum wind speed in each wind direction was analyzed, such as Fig.10 As shown in part (a), Fig.10 Part (b) shows the extreme wind speeds of each wind direction under a 100-year return period. In contrast, the typhoon extreme wind speed exceeds the constant wind speed in all wind directions, indicating that the wind climate in the area is mainly affected by typhoons. In addition, the analysis of the extreme wind speeds in different wind directions shows that D2 is the most unfavorable direction, with an extreme wind speed of 41.4 m / s. In this specific embodiment, the design parameter set includes the dimensions of the tower body's vertical rods, diagonal rods, and horizontal rods, the wall thickness and dimensions of the boom's upper chord and lower chord, the boom's lower chord dimensions, the cable dimensions, and the cable diameter; When establishing the finite element model of the tower crane, the beam unit and truss unit of the tower crane are mainly used to construct the finite element model of the tower crane. For example, the beam unit is used to equivalent the solid parts such as the slewing mechanism, and the influence of the trolley, hook, tower ladder and electrical mechanism on the dynamic characteristics of the tower crane is ignored; however, if these parts are ignored, the mass distribution of the established finite element model of the tower crane will be different from the actual structure, which affects the dynamic response accuracy of the tower crane under wind load. Therefore, it is necessary to calibrate the finite element model of the tower crane so that the dynamic characteristics of the finite element model of the tower crane match the dynamic characteristics of the tower crane at the construction site, and then proceed to step 3; Step 3. Perform sensitivity analysis on the parameter objects in the design parameter set, and determine the parameter objects that need to be calibrated in the tower crane finite element model based on the sensitivity analysis results; the calculation formula for sensitivity analysis is:

[0024] In the formula, represents the parameter sensitivity, It is the parameter object in the tower crane design parameters. is the parameter value change of the parameter object in the tower crane design parameters, is the output of the tower crane finite element model, It is the output change of the parameter object in the tower crane finite element model before and after the change; Based on the sensitivity analysis results, the parameter with the largest sensitivity value among the design parameters of the tower crane is determined as the parameter object that needs to be calibrated. The parameter objects that need to be calibrated include the tower body upright size, the wall thickness of the upper chord of the crane arm, the size of the lower chord of the crane arm and the diameter of the cable. The corresponding parameters in the finite element model of the tower crane are calibrated so that the finite element model of the tower crane has a similar mass distribution to the tower crane at the construction site, and then the modal frequency and vibration shape of the finite element model of the tower crane are consistent with the modal frequency and vibration shape of the tower crane at the construction site. Step 3 is a specific judgment method for matching the modal frequency of the calibrated finite element model of the tower crane with the modal frequency of the tower crane at the construction site.

[0025] Step 4. Calibrate the tower crane finite element model based on the parameter object that needs to be calibrated, and perform modal analysis on the calibrated finite element model to obtain the modal frequency and vibration mode of the calibrated tower crane finite element model; determine whether the vibration mode of the tower crane finite element model and the tower crane in the same mode is consistent, and whether the difference between the modal frequency of the tower crane and the modal frequency of the tower crane finite element model is less than a preset threshold value K. If so, proceed to step 5; if not, return to step 4; This specific embodiment specifically shows the first three vibration modes of the tower crane or the tower crane finite element model, wherein the first vibration mode is that the boom rotates around the tower body in the xy plane; the second vibration mode is that the tower body swings back and forth, and the boom bends and deforms; the third vibration mode is that the tower body tilts slightly, and the boom is torsionally deformed; Figure 6 , Figure 7 and 8 As shown in the figure, the first three vibration modes of the finite element model of the tower crane show that the vibration modes of the tower crane are mainly caused by the deformation of the boom, followed by the deformation of the tower body, which is consistent with the actual vibration characteristics of the tower crane.

[0026] The vibration mode and modal frequency of the finite element model of the tower crane after calibration in this specific embodiment and the vibration mode and modal frequency of the tower crane at the construction site are shown in Table 1: Table 1 Modal frequencies of the tower crane before and after finite element model calibration

[0027] It can be seen from Table 1 that the first three frequencies of the calibrated tower crane finite element model are closer to the measured frequencies of the tower crane at the construction site.

[0028] Step 5. Calculate the wind load time history of each simulation node based on the simulated wind speed, apply the wind load time history of each simulation node to the calibrated tower crane finite element model in the most unfavorable direction to obtain displacement data, and form wind speed-displacement. The simulated wind load is applied to the corresponding simulation node in a direction perpendicular to the boom (i.e., in a direction perpendicular to the boom); Thus, based on the simulated wind speed set, a data set of R groups of wind speed-displacement containing continuous time periods is obtained; The calculation process of the wind load time history of each simulation node based on the simulated wind speed extreme in the simulation operation of this specific embodiment includes: Firstly, the wind speed time history of each simulation node is calculated by the harmonic superposition method for the simulated wind speed; then, the wind load time history of each simulation node is calculated by the wind load formula for the wind speed time history of each simulation node; the tower crane finite element model is simulated based on the wind load time history of each simulation node to obtain the displacement time history of each simulation node; the maximum value in the displacement time history is selected as the displacement data of the tower crane finite element model under extreme wind speed to form the wind speed-displacement; here, the harmonic superposition method and the wind load formula are both existing technologies and will not be elaborated here.

[0029] Step 6. Perform exponential fitting on the data set obtained in step 5 to obtain a simulated data fitting line; select the construction site wind speed and tower crane displacement in the non-working state before the typhoon from the monitoring system, and perform exponential fitting to obtain the measured data fitting line; determine whether the difference between the measured data fitting line and the simulated data fitting line in the tower crane displacement at the same wind speed is less than a preset threshold value O. The preset threshold value O in this specific embodiment is less than 10 mm. If so, the calibrated tower crane finite element model matches the tower crane at the construction site, and enter step 7. If not, return to step 4; In this specific embodiment, the measured data, the corresponding measured data fitting line and the simulated data fitting line are selected from the monitoring system as shown in FIG. Fig. 9 As shown in the figure, it can be seen that the trends of the fitting lines are consistent. In addition, the figure also points out the displacement data corresponding to the two fitting lines at wind speeds of 5m / s, 10m / s, and 15m / s, where the maximum displacement difference is 6.03mm, indicating that the updated model can more accurately reflect the actual force and deformation.

[0030] Step 7. Construct a CNN-BiLSTM-Adaboost displacement response prediction model; the CNN-BiLSTM-Adaboost structural response prediction model integrates a CNN model, a BiLSTM model, and an Adaboost algorithm.

[0031] The CNN model is used to extract image features from wind speed-displacement, and output the extracted image features to the BiLSTM model. The CNN model includes an input layer, a convolution layer, a pooling layer, a fully connected layer and an output layer connected in sequence. The convolution layer is used to capture the local features of the input data and form a feature map. The convolution layer is expressed as: ; In the formula, represents the convolutional layer Output feature map, represents a nonlinear activation function, is the weight matrix, Input features to the convolutional layer, represents the deviation vector; The pooling layer is used to perform downsampling operations to reduce the dimension of the data; the pooling layer is expressed as: ; In the formula, represents the output of the pooling layer, Represents the bias of the pooling layer; The fully connected layer is used to integrate and transform the features extracted by the convolution layer and the pooling layer, and perform the final classification and regression tasks to output the prediction results; the fully connected layer is expressed as: ; In the formula, Represents the final output result, represents the weight matrix, represents the bias of the fully connected layer.

[0032] The BiLSTM model is used to perform sequence modeling on the image features. The BiLSTM model includes an input layer, a forward LSTM model, a reverse LSTM model and an output layer. The data input from the input layer to the BiLSTM model is captured through the two forward LSTM models and the reverse LSTM model, and the sequence dynamics of the image features are output from the output layer. The BiLSTM model is expressed as: ; ; ; In the formula, Indicates that the forward LSTM model is in 1~ n +1 hidden layer state at time; Indicates that the reverse LSTM model is in 1~ n +1 hidden layer state at time; Represents the inter-layer activation function; Indicates 1~n Input data at time, represents the weight between the input layer and the forward LSTM model, represents the weight between the input layer and the reverse LSTM model, represents the weight of the forward LSTM model, represents the weight of the reverse LSTM model, represents the weights of the reverse LSTM model and the output layer, Represents the weights of the forward LSTM model and the output layer; Indicates 1~ n Output data at the moment.

[0033] The Adaboost algorithm constructs a strong classifier by integrating multiple weak classifiers, and outputs the prediction result as the displacement of the tower crane under the extreme wind speed of the future typhoon by iteratively adjusting the weights of the weak classifiers; the Adaboost algorithm is expressed as: ; ; ; ; In the formula, express Group training set data, is expressed as the normalization factor, represents the input vector and the output vector, represents a weak classifier, express The weight of represents the loss error, represents the sample weight, Represents a strong classifier. In each iteration, the weak classifier Weight According to its loss error To calculate, then, the weak classifier Performance on training set data, sample weights Update, and finally, the weak classifier Ensemble into a strong classifier middle.

[0034] Step 8. Use the data set obtained in step 5 as the training set, the wind speed-displacement of the current period as the input, and the displacement of the next period as the output to train the CNN-BiLSTM-Adaboost displacement response prediction model; obtain V groups of wind speed-displacement composition test sets of continuous periods under typhoon conditions from the monitoring system, and test the trained CNN-BiLSTM-Adaboost displacement response prediction model based on the test set to realize the response prediction of tower crane displacement under typhoon conditions.

[0035] Prediction of the displacement response of a tower crane under a typhoon using the fusion data and model of this application: Dataset: A total of 1,150 data sets were constructed. The extreme wind speed and crane displacement data obtained from the finite element simulation of the crane in steps 5 and 6 were used as training sets, totaling 1,035 sets (accounting for 90% of the data set), and the test set was the wind speed and crane displacement data under typhoon conditions actually monitored by the crane selected by the monitoring platform, totaling 115 sets (accounting for 10% of the data set); Training object: CNN-BiLSTM-Adaboost displacement response prediction model; Prediction results: Fig.11 (ab) It can be seen that the prediction of the displacement response of the tower crane under the typhoon by the fusion data and model used in this application has significantly improved the accuracy of the tower top displacement prediction under typhoon conditions and the robustness of the algorithm; the RMSE of the test set is 11.7857mm; Fig.11 (cd) shows that there is a strong correlation between the measured value and the predicted value, the R2 value of the training set is 0.9963, and the R2 value of the test set is 0.9871; Fig.11 (e) It can be seen that the displacement prediction error is small and remains within 40 mm. The prediction error results of the first four samples with larger measured displacements in the test set (sample numbers are 57, 52, 82, and 90) are listed, as shown in Table 2: Table 2 Error results of the test set (4 maximum displacements under typhoon)

[0036] The maximum error of the four sets of data is controlled at about 10 mm, indicating that the CNN-BiLSTM-Adaboost displacement response prediction model can maintain a small prediction error even on samples with large measured displacements.

[0037] In summary, the data model hybrid driven method is used to predict the tower top displacement response under typhoon conditions, which effectively solves the problem of too few extreme samples in the training set and improves the accuracy and reliability of the "grey swan" event prediction.

[0038] In the description of the embodiments of the present application, it should be noted that in the description of the present application, terms such as "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description, and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present application.

[0039] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "in the present embodiment", "specific example", or "some examples" etc. means that the specific features, mechanisms, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0040] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model, wherein the tower crane comprises a tower body, a balancing arm, a lifting arm, a tie rod connecting the tower top and the balancing arm, and a tie rod connecting the tower top and the lifting arm, and the wind speed environment faced by the tower crane comprises a typhoon condition and a benign wind condition, and is characterized in that: The prediction method of tower crane displacement response under typhoon by integrating data and models includes: Step 1. Construct a monitoring system for real-time monitoring of the wind speed and displacement data of the tower crane at the construction site, wherein the monitoring system includes a plurality of inclination sensors arranged along the height direction of the tower body, an acceleration sensor arranged at the junction of the tower body and the crane arm, and an anemometer arranged at the construction site, wherein the inclination sensor is used to collect inclination data to calculate the displacement data of the tower crane, and the anemometer is used to collect the wind speed at the construction site; modal analysis is performed on the acceleration data collected by the acceleration sensor to obtain the modal frequency and vibration type of the tower crane; Step 2. Obtain the design parameter set of the tower crane, and construct a finite element model of the tower crane based on the design parameter set; divide the finite element model of the tower crane into N simulation areas on average, and determine the simulation nodes of each simulation area; and calculate the extreme wind speed under the most unfavorable wind direction of the construction site based on the typhoon full path simulation and wind field simulation as the upper limit of the simulated wind speed; randomly sample R simulated wind speeds in the range of 0 to extreme wind speed to form a wind speed sequence of continuous time periods to form a simulated wind speed set; Step 3. Perform sensitivity analysis on the parameter objects in the design parameter set, and determine the parameter objects that need to be calibrated in the tower crane finite element model according to the sensitivity analysis results; Step 4. Calibrate the tower crane finite element model based on the parameter object that needs to be calibrated, and perform modal analysis on the calibrated finite element model to obtain the modal frequency and vibration mode of the calibrated tower crane finite element model; determine whether the vibration mode of the tower crane finite element model and the tower crane in the same mode is consistent, and whether the difference between the modal frequency of the tower crane and the modal frequency of the tower crane finite element model is less than a preset threshold value K. If so, proceed to step 5; if not, return to step 4; Step 5. Calculate the wind load time history of each simulation node based on the simulated wind speed, apply the wind load time history of each simulation node to the calibrated tower crane finite element model along the most unfavorable direction to obtain displacement data, and form wind speed-displacement data. Thus, based on the simulated wind speed set, obtain R groups of wind speed-displacement data sets containing continuous time periods; Step 6. Perform exponential fitting on the data set obtained in step 5 to obtain a simulated data fitting line; obtain the wind speed and tower crane displacement of the construction site in the non-working state before the typhoon from the monitoring system, and perform exponential fitting to obtain the measured data fitting line; determine whether the difference between the measured data fitting line and the simulated data fitting line in the tower crane displacement at the same wind speed is less than a preset threshold value O. If so, the calibrated tower crane finite element model matches the tower crane at the construction site and proceeds to step 7; if not, return to step 4; Step 7. Construct a CNN-BiLSTM-Adaboost displacement response prediction model; Step 8. Use the data set obtained in step 5 as the training set, use the wind speed-displacement of the current period as input, and use the displacement of the next period as output to train the CNN-BiLSTM-Adaboost displacement response prediction model; input the training set into the CNN-BiLSTM-Adaboost displacement response prediction model for training, obtain V groups of wind speed-displacement composition test sets of continuous periods under typhoon conditions from the monitoring system, and test the trained CNN-BiLSTM-Adaboost displacement response prediction model based on the test set to realize the response prediction of tower crane displacement under typhoon conditions.

2. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 1 is characterized in that: The monitoring system constructed in step 1 includes a perception layer, a network layer and an application layer which are communicatively connected in sequence. The perception layer includes an inclinometer, an anemometer and an acceleration sensor. The anemometer is used to collect wind direction data and wind speed data at the construction site in real time. The network layer includes a GPRS network for wireless transmission of data. The application layer is composed of a remote monitoring cloud platform for displaying real-time data collected by the perception layer.

3. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 1 is characterized in that: In step 1, the multiple inclination sensors arranged along the height direction of the tower body divide the tower body into multiple monitoring segments, and each of the inclination sensors collects the three-axis inclination of the corresponding monitoring segment on the X-axis, Y-axis, and Z-axis.

4. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 3 is characterized in that: The monitoring system calculates the displacement data of the tower crane at the construction site by: First, the horizontal plane inclination angles of the inclination sensor along the X-axis and the Y-axis are obtained in turn; Next, combined with the height position of each inclination sensor on the tower body, the coordinates of each inclination sensor are calculated through the conversion formula, thereby obtaining the horizontal offset of the tower body position where each inclination sensor is located; the conversion formula is: ； In the formula, is the height position of each inclination sensor, are the horizontal plane inclination angles of the X-axis and Y-axis of each inclination sensor respectively; Finally, the horizontal offset of each tilt sensor is added in turn to obtain the total horizontal offset of the tower top as the displacement data of the tower crane.

5. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 1 is characterized in that: The calculation formula for performing sensitivity analysis on the design parameter set in step 3 is: ； In the formula, represents the parameter sensitivity, It is the parameter object in the tower crane design parameters. is the parameter value change of the parameter object in the tower crane design parameter set, is the output of the tower crane finite element model, It is the output change of the parameter object in the tower crane finite element model before and after the change; Based on the sensitivity analysis results, the parameter with the largest sensitivity value in the design parameter set of the tower crane is determined as the parameter object that needs to be calibrated, and the corresponding parameters in the finite element model of the tower crane are calibrated so that the modal frequency of the finite element model of the tower crane under the same vibration mode matches the modal frequency of the tower crane at the construction site.

6. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 1 is characterized in that: The CNN-BiLSTM-Adaboost structural response prediction model constructed in step 7 integrates the CNN model, the BiLSTM model and the Adaboost algorithm.

7. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 6 is characterized in that: The CNN model is used to extract image features from wind speed-displacement, and output the extracted image features to the BiLSTM model. The CNN model includes an input layer, a convolution layer, a pooling layer, a fully connected layer and an output layer connected in sequence. The convolution layer is used to capture the local features of the input data and form a feature map. The convolution layer is expressed as: ； In the formula, represents the convolutional layer Output feature maps, represents a nonlinear activation function, is the weight matrix, Input features to the convolutional layer, represents the deviation vector; The pooling layer is used to perform downsampling operations to reduce the dimension of the data; the pooling layer is expressed as: ； In the formula, represents the output of the pooling layer, Represents the bias of the pooling layer; The fully connected layer is used to integrate and transform the features extracted by the convolution layer and the pooling layer, and perform the final classification and regression tasks to output the prediction results; the fully connected layer is expressed as: ； In the formula, Represents the final output result, represents the weight matrix, represents the bias of the fully connected layer.

8. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 6 is characterized in that: The BiLSTM model is used to perform sequence modeling on the image features. The BiLSTM model includes an input layer, a forward LSTM model, a reverse LSTM model and an output layer. The data input from the input layer to the BiLSTM model is captured through the two forward LSTM models and the reverse LSTM model, and the sequence dynamics of the image features are output from the output layer. The BiLSTM model is expressed as: ； ； ； In the formula, Indicates that the forward LSTM model is in 1~ n +1 hidden layer state at time; Indicates that the reverse LSTM model is in 1~ n +1 hidden layer state at time; Represents the inter-layer activation function; Indicates 1~ n Input data at time, represents the weight between the input layer and the forward LSTM model, represents the weight between the input layer and the reverse LSTM model, represents the weight of the forward LSTM model, represents the weight of the reverse LSTM model, represents the weights of the reverse LSTM model and the output layer, Represents the weights of the forward LSTM model and the output layer; Indicates 1~ n Output data at that moment.

9. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 6, characterized in that: The Adaboost algorithm constructs a strong classifier by integrating multiple weak classifiers, and outputs the prediction result as the displacement of the tower crane under the extreme wind speed of the future typhoon by iteratively adjusting the weights of the weak classifiers; the Adaboost algorithm is expressed as: ； ； ； ； In the formula, express Group training set data, is expressed as the normalization factor, represents the input vector and the output vector, represents a weak classifier, express The weight of represents the loss error, represents the sample weight, Represents a strong classifier. In each iteration, the weak classifier Weight According to its loss error To calculate, then, the weak classifier Performance on training set data, sample weights Update, and finally, the weak classifier Ensemble into a strong classifier middle.

10. The method for predicting the displacement response of a tower crane under a typhoon by integrating data and a model according to claim 1, characterized in that: Calculating the wind load time history of each simulation node based on the simulated wind speed in step 5 specifically includes: first calculating the wind speed time history of each simulation node using the harmonic superposition method for the simulated wind speed; and then calculating the wind load time history of each simulation node using the wind load formula based on the wind speed time history of each simulation node.

Images