Method for detecting stress state of horizontal pushing type roof structure

Abstract

The invention discloses a method for detecting the stress state of a horizontal pushing type roof structure. The method comprises the following steps: step 1, building roof stress detection equipment; 2, roof stress detection equipment is installed; 3, designing a filter circuit, and preprocessing the collected electromagnetic signals; 4, establishing a network model of electromagnetic signal and stress mapping, wherein the network model is composed of a pooling layer, a mask multi-head self-attention network, a normalization layer, a residual connection layer, a feed-forward network, a normalization layer, a residual connection layer and a feed-forward network; 5, training a network model of electromagnetic signal and stress mapping; and step 6, evaluating the stress of the roof structure by using the trained network model. According to the network model, by combining the advantages of a plurality of components, accurate detection and evaluation of the stress state of a complex structure are realized, the stress state change of the structure is monitored in real time, and timely data support is provided for structure safety monitoring.

Description

Technical Field

[0001] The invention relates to the field of stress detection, and in particular to a method for detecting the stress state of a horizontally sliding roof structure. Background Art

[0002] As an advanced architectural form, the horizontal sliding roof structure is widely used in large public buildings such as gymnasiums and exhibition halls. Its characteristic is that the roof can slide or unfold in the horizontal direction to meet different usage requirements. Before the 1980s, the horizontal sliding roof structure was mainly opened and closed in the form of folds of membrane structure, but this opening method has inherent defects. During use, it is often affected by wind and rain, causing the opening and closing operation to fail or the membrane material to be torn. In the 1980s, with the improvement of the level of construction technology, it has become possible to build large opening and closing structures. During this period, the horizontal sliding roof building using rigid steel structures such as arched steel structures, arch shells, partial spherical shells or flat grids as load-bearing structures has developed rapidly, and its roofing materials are mostly membrane materials, metal plates and other lightweight materials. The roof system is divided into several unit pieces. Through the movement and rotation of the unit pieces, the pieces are overlapped and stacked to realize the opening and closing of the roof. Its representative buildings include: Toronto Sky Dome Stadium in Canada, Japan Ocean Dome, Fukuoka Stadium in Japan, etc. However, this kind of structure will undergo complex mechanical changes during movement, such as internal force redistribution and stress concentration, so it is particularly important to monitor and evaluate the stress state of the structure in real time.

[0003] In view of the particularity of the horizontal sliding roof structure, the present invention proposes a stress state detection method for the horizontal sliding roof structure. This method combines multiple steps such as electromagnetic signal acquisition, signal preprocessing, network model establishment and training, and stress assessment, aiming to achieve accurate detection of the stress state of the roof structure. This method can be widely used in stress state detection of various complex structures, providing strong technical support for safety monitoring and assessment in the field of structural engineering. At the same time, with the continuous development of sensor technology and machine learning algorithms, the detection accuracy and real-time performance of this method will be further improved, providing a more solid guarantee for the safety and reliability of structural engineering. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method for detecting the stress state of a horizontally sliding roof structure. In practical applications, the collected electromagnetic signals are input into a trained network model, and the model will automatically calculate the stress state of the roof structure. The evaluation results can be used to guide the safety monitoring, maintenance and management of the structure.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for detecting the stress state of a horizontal sliding roof structure comprises the following steps:

[0007] Step 1, build a roof stress detection device, which consists of a U-shaped magnetic yoke, an excitation coil, a detection coil and an A / D sampling circuit;

[0008] Step 2: Install the roof stress detection equipment in the selected detection area and debug it to ensure that the equipment can work normally and accurately collect data;

[0009] Step 3, designing a filtering circuit to pre-process the collected electromagnetic signal, including steps such as denoising and filtering, to improve the signal quality;

[0010] Step 4, establishing a network model of electromagnetic signal and stress mapping, which consists of a pooling layer, a masked multi-head self-attention network, a normalization layer, a residual connection layer, a feedforward network, a normalization layer, a residual connection layer, and a feedforward network;

[0011] Step 5, training the network model of electromagnetic signal and stress mapping, using the data set composed of signals and stresses to train the network model of electromagnetic signal and stress mapping;

[0012] Step 6: Use the trained network model to evaluate the stress of the roof structure.

[0013] Further, the process of installing the roof stress detection equipment in step 2 is expressed as:

[0014] The roof truss is in the form of a plane truss, with steel columns and diagonal braces as web members, and main beams as upper and lower chords. All components of the truss are box-shaped, and the cross-sectional material is Q355B. The stress detection system applies an alternating voltage signal to the excitation coil to magnetize the stress area of ​​the truss to be tested, and then places the U-shaped yoke and detection coil into the stress area to be tested to detect the magnetic signal in the area.

[0015] Furthermore, the process of establishing the network model of electromagnetic signal and stress mapping in step 4 is expressed as follows:

[0016] The pooling layer can reduce the dimension of the data while retaining important information. The pooling layer is selected as average pooling, and its expression is as follows:

[0017]

[0018] Where AveragePooling(x ij ) is the output of average pooling, x ij is the electromagnetic signal after processing in step 3, i and j represent the index of the signal row and column, R ijis the index set of the region, p and q are the position offsets within the region;

[0019] The masked multi-head self-attention network is used to improve the model's ability to understand the complex relationships within the data. Its expression is as follows:

[0020]

[0021] In the formula, A represents the attention weight, sofmax() is the softmax function, Q is the output of the pooling layer, K is the key, V is the value, and d k is the rank of the key;

[0022] The normalization layer can speed up the network training process and prevent the gradient from disappearing or exploding. LN normalization is used;

[0023] The residual connection layer allows the gradient to propagate directly to the deeper layers of the network, which helps to train deep networks. Its expression is as follows:

[0024] y=W2σ1(W1u)+W s u

[0025] Where y is the output of the residual connection layer, u is the input of the residual connection layer, W1 and W2 are weight matrices, σ1() is the activation function ReLU, and W s is the dimension-matched weight matrix;

[0026] The feedforward network can perform nonlinear transformation on features and enhance the expressiveness of the model;

[0027] O=σ2(∑W o y+b)

[0028] Where O is the output of the feedforward network layer, y is the input of the feedforward network layer, σ2() is the activation function Sigmoid, and W o is the feedforward network weight, and b is the feedforward network layer bias.

[0029] Furthermore, the stress assessment process in step 6 can be expressed as follows:

[0030] The trained model is integrated into the real-time monitoring system to continuously monitor the stress of the roof structure, regularly analyze the monitoring data, and generate reports to evaluate the long-term health of the roof and predict potential failures; when abnormal stress levels are detected, the system should automatically trigger an early warning to notify relevant personnel to take necessary maintenance or emergency measures. Compared with the prior art, the beneficial effects of the present invention are:

[0031] 1. The present invention does not need to directly contact the surface of the structure, thus reducing interference and damage to the structure;

[0032] 2. The network model of the present invention achieves accurate detection and evaluation of the stress state of complex structures by combining the advantages and strengths of multiple components;

[0033] 3. The combination of network models in the present invention not only improves the performance and stability of the model, but also reduces the training difficulty and calculation cost;

[0034] 4. The present invention can monitor the changes in the stress state of the structure in real time and provide timely data support for structural safety monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a truss elevation layout diagram of the present invention;

[0036] Figure 2 This is a network model diagram of the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments:

[0038] The present invention proposes a method for detecting the stress state of a horizontal sliding roof structure, wherein the truss elevation layout diagram is as follows: Figure 1 As shown in the figure, the network model is as follows Figure 2 As shown, the steps of the present invention are described in detail below.

[0039] Step 1, build a roof stress detection device, which consists of a U-shaped magnetic yoke, an excitation coil, a detection coil and an A / D sampling circuit;

[0040] Step 2: Install the roof stress detection equipment in the selected detection area and debug it to ensure that the equipment can work normally and accurately collect data;

[0041] The roof truss is a plane truss, with steel columns and diagonal braces as web members, and main beams as upper and lower chords. All truss members are box-shaped, and the cross-section material is Q355B. The truss elevation layout is shown in the figure below. Figure 1 As shown, the stress detection system applies an alternating voltage signal to the excitation coil to magnetize the stress area to be measured on the truss, and then places the U-shaped magnetic yoke and the detection coil into the stress area to be measured to detect the magnetic signal in the area.

[0042] Step 3, designing a filtering circuit to pre-process the collected electromagnetic signal, including steps such as denoising and filtering, to improve the signal quality;

[0043] Step 4: Establish a network model for electromagnetic signal and stress mapping. The network model consists of a pooling layer, a masked multi-head self-attention network, a normalization layer, a residual connection layer, a feedforward network, a normalization layer, a residual connection layer, and a feedforward network. The network model is shown in the figure below: Figure 2 As shown;

[0044] The pooling layer can reduce the dimension of the data while retaining important information. The pooling layer is selected as average pooling, and its expression is as follows:

[0045]

[0046] Where AveragePooling(x ij ) is the output of average pooling, x ij is the electromagnetic signal after processing in step 3, i and j represent the index of the signal row and column, R ij is the index set of the region, p and q are the position offsets within the region;

[0047] The masked multi-head self-attention network is used to improve the model's ability to understand the complex relationships within the data. Its expression is as follows:

[0048]

[0049] In the formula, A represents the attention weight, sofmax() is the softmax function, Q is the output of the pooling layer, K is the key, V is the value, and d k is the rank of the key;

[0050] The normalization layer can speed up the network training process and prevent the gradient from disappearing or exploding. LN normalization is used;

[0051] The residual connection layer allows the gradient to propagate directly to the deeper layers of the network, which helps to train deep networks. Its expression is as follows:

[0052] y=W2σ1(W1u)+W s u

[0053] Where y is the output of the residual connection layer, u is the input of the residual connection layer, W1 and W2 are weight matrices, σ1() is the activation function ReLU, and W s is the dimensionally matched weight matrix.

[0054] The feedforward network can perform nonlinear transformation on features and enhance the expressiveness of the model;

[0055] O=σ2(∑W o y+b)

[0056] Where O is the output of the feedforward network layer, y is the input of the feedforward network layer, σ2() is the activation function Sigmoid, and W o is the feedforward network weight, and b is the feedforward network layer bias.

[0057] Step 5, training the network model of electromagnetic signal and stress mapping, using the data set composed of signals and stresses to train the network model of electromagnetic signal and stress mapping;

[0058] Step 6: Use the trained network model to evaluate the stress of the roof structure.

[0059] Integrate the trained model into the real-time monitoring system to conduct continuous stress monitoring of the roof structure, regularly analyze the monitoring data, and generate reports to assess the long-term health of the roof and predict potential failures; when abnormal stress levels are detected, the system should automatically trigger an early warning to notify relevant personnel to take necessary maintenance or emergency measures.

[0060] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. A method for detecting the stress state of a horizontal sliding roof structure, characterized in that: The following steps are involved: Step 1, build a roof stress detection device, which consists of a U-shaped magnetic yoke, an excitation coil, a detection coil and an A / D sampling circuit; Step 2: Install the roof stress detection equipment in the selected detection area and debug it to ensure that the equipment can work normally and accurately collect data; Step 3, designing a filtering circuit to pre-process the collected electromagnetic signal, including steps such as denoising and filtering, to improve the signal quality; Step 4, establishing a network model of electromagnetic signal and stress mapping, which consists of a pooling layer, a masked multi-head self-attention network, a normalization layer, a residual connection layer, a feedforward network, a normalization layer, a residual connection layer, and a feedforward network; Step 5, training the network model of electromagnetic signal and stress mapping, using the data set composed of signals and stresses to train the network model of electromagnetic signal and stress mapping; Step 6: Use the trained network model to evaluate the stress of the roof structure.

2. A method for detecting stress state of a horizontal sliding roof structure according to claim 1, characterized in that: The process of installing the roof stress detection equipment in step 2 is expressed as: The roof truss is in the form of a plane truss, with steel columns and diagonal braces as web members, and main beams as upper and lower chords. All components of the truss are box-shaped, and the cross-sectional material is Q355B. The stress detection system applies an alternating voltage signal to the excitation coil to magnetize the stress area of ​​the truss to be tested, and then places the U-shaped yoke and detection coil into the stress area to be tested to detect the magnetic signal in the area.

3. A method for detecting stress state of a horizontal sliding roof structure according to claim 1 or 2, characterized in that: The process of establishing the network model of electromagnetic signal and stress mapping in step 4 is as follows: The pooling layer can reduce the dimension of the data while retaining important information. The pooling layer is selected as average pooling, and its expression is as follows: Where AveragePooling(x ij ) is the output of average pooling, x ij is the electromagnetic signal after processing in step 3, i and j represent the index of the signal row and column, R ij is the index set of the region, p and q are the position offsets within the region; The masked multi-head self-attention network is used to improve the model's ability to understand the complex relationships within the data. Its expression is as follows: In the formula, A represents the attention weight, sofmax() is the softmax function, Q is the output of the pooling layer, K is the key, V is the value, and d k is the rank of the key; The normalization layer can speed up the network training process and prevent the gradient from disappearing or exploding. LN normalization is used; The residual connection layer allows the gradient to propagate directly to the deeper layers of the network, which helps to train deep networks. Its expression is as follows: y=W2σ1(W1u)+W s u Where y is the output of the residual connection layer, u is the input of the residual connection layer, W1 and W2 are weight matrices, σ1() is the activation function ReLU, and W s is the dimension-matched weight matrix; The feedforward network can perform nonlinear transformation on features and enhance the expressiveness of the model; O=σ2(∑W o y+b) Where O is the output of the feedforward network layer, y is the input of the feedforward network layer, σ2() is the activation function Sigmoid, and W o is the feedforward network weight, and b is the feedforward network layer bias.

4. A method for detecting stress state of a horizontal sliding roof structure according to claim 1, characterized in that: The process of stress assessment in step 6 can be expressed as follows: Integrate the trained model into the real-time monitoring system to conduct continuous stress monitoring of the roof structure, regularly analyze the monitoring data, and generate reports to assess the long-term health of the roof and predict potential failures; when abnormal stress levels are detected, the system should automatically trigger an early warning to notify relevant personnel to take necessary maintenance or emergency measures.

Images