Ancient building durability monitoring system based on micro-deformation monitoring and machine learning

Abstract

The invention discloses an ancient building durability monitoring system based on micro-deformation monitoring and machine learning, which is used for predicting the water content of a wood structure of an ancient building through monitoring local wood grain micro-deformation of the wood structure of the ancient building and machine learning based on a multi-Fick coupling unsteady state water transfer model and Bayesian inversion, and further monitoring the durability of the ancient building. According to the method, digital image correlation is used, the sub-pixel precision is considered, and the machine learning model is corrected through temperature and humidity, so that the durability of the ancient building is monitored more accurately.

Description

Technical Field

[0001] The present invention belongs to the technical field of ancient building health monitoring, and in particular relates to an ancient building durability monitoring system based on micro-deformation monitoring and machine learning. Background Art

[0002] There are a large number of ancient buildings in my country, most of which are wooden structures. The moisture content of wooden structures significantly affects the stiffness and strength of wooden structures, as well as mechanical properties such as shrinkage and swelling, mechanical moisture creep, etc. Due to the influence of natural and social factors, such as earthquake damage, wind and rain erosion, biological invasion and crowd interference, the structural material performance of ancient buildings has gradually degraded, the bearing capacity has decreased, and different degrees and types of damage have occurred. Existing computer vision-based health monitoring of ancient buildings mainly focuses on the deformation or displacement monitoring of building structures, lacks local monitoring of weak parts such as load-bearing components, and the monitoring results are inaccurate and incomplete. Summary of the invention

[0003] In response to the above technical problems, the present invention proposes an ancient building durability monitoring system based on micro-deformation monitoring and machine learning. By monitoring the local wood grain micro-deformation of the ancient building wooden structure and machine learning based on the Dophick coupling non-steady-state water transfer model and Bayesian inversion, the moisture content of the ancient building wooden structure is predicted, thereby monitoring the durability of the ancient building.

[0004] In order to achieve the above technical objectives, an ancient building durability monitoring system based on micro-deformation monitoring and machine learning is provided, wherein the ancient building durability monitoring system based on micro-deformation monitoring and machine learning comprises: A feature extraction module, connected to the industrial computer module, is used to measure the wood grain and extract feature points based on visual recognition; The deformation monitoring module is connected to the industrial computer module and is used to monitor the slight deformation of the wooden structure based on digital image correlation and considering sub-pixel accuracy; The durability monitoring module is connected to the industrial computer module and is used to use machine learning to output the moisture content of the wood structure and then monitor the durability of the wood structure.

[0005] Furthermore, the ancient building durability monitoring system based on micro-deformation monitoring and machine learning includes: The video monitoring module is connected to the industrial computer module and is used to perform real-time video monitoring of the ancient building using the camera equipment set at the video monitoring point of the ancient building; The temperature and humidity monitoring module is connected to the industrial computer module and is used to collect the temperature and humidity data of the ancient building using sensors set at the temperature and humidity monitoring points of the ancient building; The industrial computer module is connected with the feature extraction module, the deformation monitoring module, the durability monitoring module, the video monitoring module, the temperature and humidity monitoring module, and the storage module, and is used to control the normal operation of each module using a small computer; The storage module is connected to the industrial computer module and is used to store the collected relevant data and various monitoring results.

[0006] Furthermore, the wood grain is measured by visually identifying the wood grain and the calibration image in the same plane, separating the wood grain area and establishing a mapping between the wood grain pixel size and the real size, and the wood grain pixel orientation and the real orientation; The feature point extraction is to extract the observed wood grain surface feature points through a feature extraction function.

[0007] Furthermore, the ancient building durability monitoring system based on micro-deformation monitoring and machine learning is characterized in that the digital image correlation and consideration of sub-pixel accuracy utilizes sub-pixel accuracy to amplify the deformation of characteristic points on the wood grain surface in the frequency domain and measure their displacement. Experiments have shown that the accuracy can reach 1mm floating error within 30m.

[0008] Furthermore, the machine learning is based on massive Dophick process simulations to construct the law of non-steady-state water transfer inside wooden structures; There is a complex water transfer process in wood structures. Water transfer causes the moisture content in different parts of the wood structure to change, which in turn generates micro-strains and internal stresses, leading to warping and cracking of the wood structure. The fluctuation of the moisture content of wood under stress will cause mechanical hygroscopic creep, which is usually the main cause of long-term deformation of wood structures. By measuring the surface strain of the wood structure, the internal moisture content of the wood structure is obtained through the law of unsteady-state water transfer inside the wood structure and Bayesian inversion.

[0009] The Bayesian inversion utilizes the Bayesian method to invert key parameters, realize micro-deformation, temperature and humidity input, output the moisture content of the wood structure, and then monitor the durability of the wood structure.

[0010] Furthermore, the temperature and humidity monitoring module collects temperature and humidity data of the ancient building using sensors arranged at the temperature and humidity monitoring points of the ancient building, including: (1) Using temperature and humidity sensors to collect ambient temperature and humidity data, the data were used to modify the Dophick coupling non-steady-state water transfer model and improve the accuracy of machine learning output of the moisture content of the wood structure; (2) Use temperature sensors installed at the temperature and humidity monitoring points of ancient buildings to collect temperature data of the observation area inside the ancient buildings; (3) Use humidity sensors installed at the temperature and humidity monitoring points of ancient buildings to collect humidity data in the observation area inside the ancient buildings; Furthermore, the small computer includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor implements the ancient building durability monitoring system based on micro-deformation monitoring and machine learning as described in any one of claims 1-7.

[0011] Beneficial Effects In combination with the above technical solutions and the technical problems solved, the technical solution of the present invention has the following beneficial effects: First, the existing technology focuses on monitoring the deformation and displacement of ancient building structures. The present invention realizes complex internal humidity deduction by monitoring surface features and tiny deformations, making the durability monitoring of ancient buildings more accurate and specific.

[0012] Second, using mature digital image correlation algorithms and considering sub-pixel accuracy to analyze two-dimensional images, micron-level micro-deformation monitoring and 10 -8 Level strain monitoring ensures accurate analysis results.

[0013] Third, use machine learning methods to construct the laws of complex non-steady-state water transfer and surface strain evolution to improve efficiency.

[0014] Fourthly, the device and composition of the present invention are simple, and the designed algorithm can be written into a separate program, which can monitor and analyze the durability of ancient buildings in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the composition of an ancient building durability monitoring system based on micro-deformation monitoring and machine learning provided by an example of the present invention; Figure 2 This is a workflow diagram of an ancient building durability monitoring system based on micro-deformation monitoring and machine learning provided by an example of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Example

[0017] The present invention proposes an ancient building durability monitoring system based on micro-deformation monitoring and machine learning. By monitoring the local wood grain micro-deformation of the ancient building wooden structure and machine learning based on the Dophick coupling non-steady-state water transfer model and Bayesian inversion, the moisture content of the ancient building wooden structure is predicted, thereby monitoring the durability of the ancient building.

[0018] like Figure 1As shown, an ancient building durability monitoring system based on micro-deformation monitoring and machine learning, the ancient building durability monitoring system based on micro-deformation monitoring and machine learning includes: The video monitoring module is connected to the industrial computer module and is used to perform real-time video monitoring of the ancient building using the camera equipment set at the video monitoring point of the ancient building; The temperature and humidity monitoring module is connected to the industrial computer module and is used to collect the temperature and humidity data of the ancient building using sensors set at the temperature and humidity monitoring points of the ancient building; A feature extraction module, connected to the industrial computer module, is used to measure the wood grain and extract feature points based on visual recognition; The wood grain measurement is to separate the wood grain area and establish a mapping between the wood grain pixel size and the real size, and the wood grain pixel orientation and the real orientation by visually identifying the wood grain and the calibration image in the same plane; The feature point extraction is to extract the observed wood grain surface feature points through a feature extraction function.

[0019] The deformation monitoring module is connected to the industrial computer module and is used to monitor the slight deformation of the wooden structure based on digital image correlation and considering sub-pixel accuracy; The ancient building durability monitoring system based on micro-deformation monitoring and machine learning is characterized in that the digital image correlation and consideration of sub-pixel accuracy utilize sub-pixel accuracy to amplify the deformation of characteristic points on the wood grain surface in the frequency domain and measure their displacement. Experiments have shown that the accuracy can reach 1mm floating error within 30m.

[0020] The durability monitoring module is connected to the industrial computer module and is used to use machine learning to output the moisture content of the wood structure and then monitor the durability of the wood structure.

[0021] The machine learning is based on massive Dophick process simulation to construct the law of non-steady-state water transfer inside the wood structure; There is a complex water transfer process in wood structures. Water transfer causes the moisture content in different parts of the wood structure to change, which in turn generates micro-strains and internal stresses, leading to warping and cracking of the wood structure. The fluctuation of the moisture content of wood under stress will cause mechanical hygroscopic creep, which is usually the main cause of long-term deformation of wood structures. By measuring the surface strain of the wood structure, the internal moisture content of the wood structure is obtained through the law of unsteady-state water transfer inside the wood structure and Bayesian inversion.

[0022] The Bayesian inversion utilizes the Bayesian method to invert key parameters, realize micro-deformation, temperature and humidity input, output the moisture content of the wood structure, and then monitor the durability of the wood structure.

[0023] like Figure 2As shown, the workflow of an ancient building durability monitoring system based on micro-deformation monitoring and machine learning provided by the example of the present invention includes: S1-1, using the video equipment installed at the video monitoring point of the ancient building to conduct real-time video monitoring of the ancient building; S1-2, measuring wood grain and extracting feature points based on visual recognition; S1-3, monitoring of small deformations of wooden structures based on digital image correlation and considering sub-pixel accuracy; S2-1, collecting temperature and humidity data of the ancient building using sensors installed at the temperature and humidity monitoring points of the ancient building; S3-1, using machine learning to output the moisture content of wood structures; S3-2, monitoring the durability of wooden structures.

Claims

1. An ancient building durability monitoring system based on micro-deformation monitoring and machine learning, characterized in that: The ancient building durability monitoring system based on micro-deformation monitoring and machine learning includes: A feature extraction module, connected to the industrial computer module, is used to measure the wood grain and extract feature points based on visual recognition; The deformation monitoring module is connected to the industrial computer module and is used to monitor the slight deformation of the wooden structure based on digital image correlation and considering sub-pixel accuracy; The durability monitoring module is connected to the industrial computer module and is used to use machine learning to output the moisture content of the wood structure and then monitor the durability of the wood structure.

2. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 1 is characterized in that: The ancient building durability monitoring system based on micro-deformation monitoring and machine learning also includes: The video monitoring module is connected to the industrial computer module and is used to perform real-time video monitoring of the ancient building using the camera equipment set at the video monitoring point of the ancient building; The temperature and humidity monitoring module is connected to the industrial computer module and is used to collect the temperature and humidity data of the ancient building using sensors set at the temperature and humidity monitoring points of the ancient building; The industrial computer module is connected with the feature extraction module, the deformation monitoring module, the durability monitoring module, the video monitoring module, the temperature and humidity monitoring module, and the storage module, and is used to control the normal operation of each module using a small computer; The storage module is connected to the industrial computer module and is used to store the collected relevant data and various monitoring results.

3. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 1 is characterized in that: The wood grain measurement is to separate the wood grain area and establish a mapping between the wood grain pixel size and the real size, and the wood grain pixel orientation and the real orientation by visually identifying the wood grain and the calibration image in the same plane; The feature point extraction is to extract the observed wood grain surface feature points through a feature extraction function.

4. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 1 is characterized in that: The digital image correlation and sub-pixel precision consideration utilizes sub-pixel precision to amplify the deformation of characteristic points on the wood grain surface in the frequency domain and measure their displacement. Experiments have proven that the accuracy can reach 1mm floating error within 30m.

5. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 1 is characterized in that: The machine learning is based on massive Dophick process simulation to construct the law of non-steady-state water transfer inside the wood structure; There is a complex water transfer process in wood structures. Water transfer causes the moisture content in different parts of the wood structure to change, which in turn generates micro-strains and internal stresses, leading to warping and cracking of the wood structure. The fluctuation of the moisture content of wood under stress will cause mechanical hygroscopic creep, which is usually the main cause of long-term deformation of wood structures. By measuring the surface strain of the wood structure, the internal moisture content of the wood structure is obtained through the law of unsteady-state water transfer inside the wood structure and Bayesian inversion.

6. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 5 is characterized in that: The Bayesian inversion utilizes the Bayesian method to invert key parameters, realize micro-deformation, temperature and humidity input, output the moisture content of the wood structure, and then monitor the durability of the wood structure.

7. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 2 is characterized in that: The temperature and humidity monitoring module collects temperature and humidity data of the ancient building using sensors arranged at the temperature and humidity monitoring points of the ancient building, including: (1) Using temperature and humidity sensors to collect ambient temperature and humidity data, the data were used to modify the Dophick coupling non-steady-state water transfer model and improve the accuracy of machine learning output of the moisture content of the wood structure; (2) Use temperature sensors installed at the temperature and humidity monitoring points of ancient buildings to collect temperature data of the observation area inside the ancient buildings; (3) Use humidity sensors installed at temperature and humidity monitoring points in ancient buildings to collect humidity data in the observation area inside the ancient buildings.

8. The ancient building durability monitoring system based on micro-deformation monitoring and machine learning according to claim 2 is characterized in that: The small computer includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor implements the ancient building durability monitoring system based on micro-deformation monitoring and machine learning as described in any one of claims 1-7.

Images