Thermal wave detection method for hole defects in plate in complex environment

Abstract

The invention discloses a thermal wave detection method for hole defects in a plate in a complex environment, and relates to a profile detection method which is realized by fusing an edge detection algorithm, heat transfer theory and deep learning. The heat transfer model is interacted with Ansys simulation software through a Python programming language and a PyAnsys library, and material attribute data is established; the deep learning model is composed of a yov5 visual recognition algorithm, and the edge detection algorithm is a Canny edge detection algorithm; firstly, an initial model of heat transfer theory is established, then a simulation model is reversely corrected through detection results of deep learning and an edge detection algorithm, and finally, the defect edge of the hole in the plate is detected. According to the method, the plate defects are detected through fusion of an edge detection algorithm and finite element heat transfer model deep learning, and the situation that hole edge detection is inaccurate due to low contrast, unclear boundary and noise area interference of a thermal image in a complex environment is avoided.

Description

Technical Field

[0001] The invention relates to a method for detecting the interior of a profile, and in particular to a thermal wave detection method for hole defects inside a plate in a complex environment. Background Art

[0002] At present, the method for detecting hole defects inside plates is usually to detect holes visible on one side, while the detection effect of blind holes inside plates is not good. Due to the irregular internal structure of the plate, which leads to the uneven distribution of the surface temperature field, the edges of hole defects in the image are usually detected using traditional edge detection algorithms to detect images under normal lighting. The current Canny, Sobel, and watershed algorithms can only detect the approximate edges of infrared images, and the detection accuracy is insufficient, so the error is large. Especially for low contrast in complex environments, unclear boundaries, and noise area interference, which cause inaccurate hole edge detection, there is an urgent need for a method that combines multiple models to improve edge detection accuracy. Summary of the invention

[0003] The purpose of the present invention is to provide a thermal wave detection method for hole defects inside plates in complex environments. The method integrates edge detection algorithm with heat transfer and deep learning models to identify the edges of hole defects inside plates, thereby realizing a material non-destructive detection method that combines thermal wave edge detection algorithm with heat transfer and computers.

[0004] The method proposed by the present invention is achieved through the following technical solutions: A thermal wave detection method for hole defects inside plates in complex environments, the method is a detection method achieved by integrating edge detection algorithm, heat transfer, and deep learning; the heat transfer model is implemented by Python programming language interacting with Ansys simulation software through PyAnsys library, and material property data is established; the deep learning model is composed of yolov5 visual recognition algorithm, and the edge detection algorithm is Canny edge detection algorithm; firstly, an initial model of heat transfer is established, and then the simulation model is reversely corrected through the detection results of deep learning and edge detection algorithm, and finally the edge of hole defects inside the plate is detected; The specific process is as follows: (1) Use Python programming language to interact with Ansys simulation software through PyAnsys library and the field environment, establish finite element heat transfer model based on material property data, and extract infrared image data; then use convolutional neural network to detect and identify hole defects inside the plate, obtain the center point position P of the hole defect and the equivalent radius R of the hole defect in the infrared image; compare the center point position error of the hole defect between the two, if it is less than the set threshold e1, the experiment can continue, otherwise the hole defect center point position P and hole defect radius R will be reversely corrected and updated to re-establish the heat transfer model, and then the experiment will be carried out in sequence; (2) The Canny operator is used to simultaneously detect the hole edges of the infrared image data collected by the thermal imaging field instrument, and the area T with a large boundary temperature gradient is identified. If the error of the hole edge area T is greater than the set threshold e2, the simulation model is re-established based on the extracted hole edge conditions and features inside the plate, and the experiments are carried out in sequence; if the features of the two are consistent, no update and correction is performed, thus avoiding inaccurate hole edge detection caused by low brightness contrast of thermal images in complex environments and unclear hole boundary features; (3) By reversely updating the heat transfer simulation model, the center point position of the hole, the hole radius, and the temperature distribution at the hole edge are corrected, and finally the internal hole defect detection results of the plate are accurately obtained.

[0005] The advantages and effects of the present invention are: 1. The present invention is a thermal wave detection method for internal hole defects in complex environment plates that combines edge detection algorithm with heat transfer and deep learning. A thermal imager is used to collect infrared images of the scene and transmit data through the network. An initial heat transfer model is first established, and then the simulation model is reversely corrected through the detection results of deep learning and edge detection algorithm to achieve the purpose of detecting hole defects inside the plate.

[0006] 2. The present invention uses Python programming language to interact with Ansys simulation software through PyAnsys library and the field environment, establishes a finite element heat transfer model based on material property data, and extracts infrared image data. Then, the hole defects inside the plate are detected and identified through a convolutional neural network, and the center point position P of the hole defect and the equivalent radius R of the hole defect in the infrared image are obtained. The center point position P error of the hole defect is compared between the two. If it is within the set threshold e1, the experiment is continued, otherwise the heat transfer model is reversely corrected.

[0007] 3. Use the traditional edge detection algorithm to detect the infrared thermal wave image and identify the area T with a large boundary temperature gradient. If the error of the hole edge area T is greater than the set threshold e2, the simulation model is re-established based on the extracted hole edge conditions and features inside the plate, and the experiments are then carried out in sequence. This method integrates the traditional edge detection algorithm with heat transfer and deep learning to avoid inaccurate hole edge detection caused by low brightness contrast of thermal images in complex environments, unclear boundaries of hole defects inside the plate, and interference from noise areas in complex environments.

[0008] 4. This method can avoid interference from noise areas in complex environments, is suitable for edge detection of infrared defects of different shapes, and improves detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a flow chart of the thermal wave edge detection method for internal hole defects in complex environment plates based on the fusion of heat transfer and deep learning of the present invention. DETAILED DESCRIPTION

[0010] The present invention will be described in detail below with reference to the embodiments shown in the accompanying drawings.

[0011] The present invention provides a thermal wave detection method for hole defects inside a plate in a complex environment, and the process is as follows: (1) This detection method and device integrates edge detection algorithm, heat transfer and deep learning to achieve the purpose of defect detection. The heat transfer model uses Python programming language, interacts with Ansys simulation software through PyAnsys library and establishes the field environment and material property data. The deep learning model is composed of yolov5 visual recognition algorithm, and the edge detection algorithm uses Canny edge detection algorithm.

[0012] (2) First, an initial heat transfer model is established. Then, the simulation model is reversely corrected through the detection results of deep learning and edge detection algorithms. The model and detection results are iteratively optimized to achieve the purpose of accurately detecting hole defects inside the plate.

[0013] (3) For the heat transfer model, the initial Python programming language interacts with the Ansys simulation software and the field environment through the PyAnsys library, the plate material property data is established, and then the heat transfer image data is extracted.

[0014] (4) For data transmission to the deep learning algorithm, a data set is first established based on the infrared thermal wave images collected on site. Then, the holes are detected and identified through a convolutional neural network to obtain the center point position P of the hole in the infrared image and the equivalent radius R of the hole.

[0015] (5) The heat transfer image data extracted from the finite element heat transfer model is compared with the results of the convolutional neural network detection and identification of the internal holes of the plate. If the error of the center point position P of the hole in both is less than the threshold e1, the Canny operator detection and identification can be performed; otherwise, the hole center point position P and the hole radius R are reversely updated to re-establish the heat transfer model, and then repeated iterations are performed for simulation optimization modeling.

[0016] (6) Then, the Canny operator detects the original infrared thermal wave image and identifies the area T with a large boundary temperature gradient. If the error of the hole edge area T is greater than the set threshold e2, the simulation model is re-established based on the extracted edge conditions and features of the hole inside the plate, and finally the edge detection result of the hole defect inside the plate is accurately obtained. Example

[0017] like Figure 1 As shown in the figure, it is a flow chart of the thermal wave detection method and device for internal hole defects of complex thermal wave plates based on the fusion of heat transfer and deep learning of the present invention, and the implementation method includes the following steps: (1) First, the thermal imaging on-site instrument data is collected, and the on-site environmental data and plate material property parameter data are transmitted through the Internet. A finite element heat transfer simulation model is established, and the simulation model image is extracted. At the same time, the collected infrared image data is transmitted to the convolutional neural network through the Internet. The convolutional neural network detects the holes inside the plate, identifies the hole area, and obtains its center point position P and equivalent radius R, thereby avoiding inaccurate hole edge detection caused by interference from noise areas in complex environments.

[0018] (2) Then compare the position errors of the center points of the hole defects of the two. If they are less than the set threshold e1, the experiment can continue. Otherwise, the position of the center point P of the hole defect and the hole radius R will be corrected and updated in the reverse direction to re-establish the heat transfer model, and then the experiment will be carried out in sequence.

[0019] (3) Then, the Canny operator is used to simultaneously detect the hole edges of the infrared image data collected by the thermal imaging field instrument, and the area T with a large boundary temperature gradient is identified. If the error of the hole edge area T is greater than the set threshold e2, the simulation model is re-established based on the extracted hole edge conditions and features inside the board, and the experiments are carried out in sequence. If the features of the two are consistent, the correction is not updated, which avoids the inaccurate hole edge detection caused by the low brightness contrast of the thermal image in a complex environment and the unclear hole boundary features. Finally, the edge detection results of the hole defects inside the wooden board are obtained.

[0020] (4) This method can avoid inaccurate edge detection of hole defects caused by low brightness contrast of thermal images, unclear hole boundaries, and interference from noise areas in complex environments through reverse correction and update design.

Claims

1. A thermal wave detection method for hole defects inside plates in complex environments, characterized in that: The method is a detection method achieved by integrating edge detection algorithm, heat transfer and deep learning; the heat transfer model is implemented by Python programming language interacting with Ansys simulation software through PyAnsys library, and material property data is established; the deep learning model is composed of yolov5 visual recognition algorithm, and the edge detection algorithm is Canny edge detection algorithm; the initial model of heat transfer is established first, and then the simulation model is reversely corrected through the detection results of deep learning and edge detection algorithm, and finally the edge of hole defects inside the plate is detected; The specific process is as follows: (1) Use Python programming language to interact with Ansys simulation software through PyAnsys library and the field environment, establish finite element heat transfer model based on material property data, and extract infrared image data; then use convolutional neural network to detect and identify hole defects inside the plate, obtain the center point position P of the hole defect and the equivalent radius R of the hole defect in the infrared image; compare the center point position error of the hole defect between the two, if it is less than the set threshold e1, the experiment can continue, otherwise the hole defect center point position P and hole defect radius R will be reversely corrected and updated to re-establish the heat transfer model, and then the experiment will be carried out in sequence; (2) The Canny operator is used to simultaneously detect the hole edges of the infrared image data collected by the thermal imaging field instrument, and the area T with a large boundary temperature gradient is identified. If the error of the hole edge area T is greater than the set threshold e2, the simulation model is re-established based on the extracted hole edge conditions and features inside the plate, and the experiments are carried out in sequence; if the features of the two are consistent, no update and correction is performed, thus avoiding inaccurate hole edge detection caused by low brightness contrast of thermal images in complex environments and unclear hole boundary features; (3) By reversely updating the heat transfer simulation model, the center point position of the hole, the hole radius, and the temperature distribution at the hole edge are corrected, and finally the internal hole defect detection results of the plate are accurately obtained.

Images