Ray detection-based intelligent detection method and system for internal defects of power capacitor

Abstract

The invention discloses an intelligent detection method and system for internal defects of a power capacitor based on ray detection. The method comprises the following steps: acquiring ray detection images of different parts and different angles of the power capacitor, and preprocessing the ray detection images; marking the preprocessed radiographic detection image, marking the position and the type of a defect in the radiographic detection image, and finally generating a data set; building a deep neural network model, and training the deep neural network model by using the data set to obtain a trained intelligent defect detection model; and utilizing the trained defect intelligent detection model to identify the type and the number of defects in the test sample detection image. The method has the advantages of nondestructive detection, high detection precision and the like.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of power capacitors, and in particular to a method and system for intelligently detecting internal defects of power capacitors based on ray detection. Background Art

[0002] Power capacitors are one of the important components in electrical equipment. During its operation, its reliability will be affected by many factors, such as vibration, current overload, and component aging. Among them, vibration is the main factor that stimulates its internal hidden defects. Internal hidden defects such as solder joint bubbles and tiny cracks will expand rapidly during the vibration process, resulting in circuit breakage, overload, short circuit and other phenomena of power capacitors, and then leading to the complete paralysis of the entire circuit system of the equipment. However, traditional testing methods including current difference and voltage techniques can only be used to detect whether the electrical performance of the capacitor meets the use standards, but cannot detect its internal hidden defects.

[0003] X-ray nondestructive testing technology was first used in military and aerospace fields. It is a cutting-edge technology for nondestructive testing. At present, X-ray nondestructive testing technology has been applied to many fields such as automobiles, chips, and aerospace to detect internal defects of key components. For a long time, manual detection methods have occupied an important position in the detection and segmentation of X-ray image defects. This method uses the naked eye to identify defects in X-ray detection images. It not only has low detection efficiency but also has strong subjectivity and poor robustness. Since the 21st century, thanks to the rapid development of computer technology, it is particularly important to use computer vision (CV) to realize the detection and segmentation of workpiece defects. However, due to the complexity of the detection and segmentation of workpiece defects, traditional CV technology cannot meet the actual requirements in terms of accuracy and operability. Summary of the invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a method and system for intelligent detection of internal defects of power capacitors based on X-ray detection with non-destructive detection and high detection accuracy.

[0005] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0006] An intelligent detection method for internal defects of a power capacitor based on X-ray detection comprises the following steps:

[0007] Obtaining radiographic inspection images of different parts and angles of the power capacitor and preprocessing the radiographic inspection images;

[0008] Mark the preprocessed X-ray detection images, mark the locations and types of defects in the X-ray detection images, and finally generate a data set;

[0009] Build a deep neural network model and use the data set to train the deep neural network model to obtain a trained defect intelligent detection model;

[0010] The trained defect intelligent detection model is used to identify the type and quantity of defects in the test sample inspection image.

[0011] Preferably, the radiation detection image is processed by adopting image filtering and enhanced image processing technology to filter out image noise.

[0012] Preferably, after the data set is generated, the data set is expanded by random cropping and free transformation.

[0013] Preferably, the deep neural network model comprises three parts: a backbone network layer, an additional feature layer and a feature classification and regression layer; wherein the backbone network layer is used to compress images and extract feature information in images; the additional feature layer is used to derive a network of multi-scale feature maps; the feature classification and regression layer is used to perform target classification and position regression on anchor frames; an attention mechanism module is added after the backbone network layer to enhance the network's ability to recognize small defects.

[0014] Preferably, the backbone network layer adopts a MobileNet, VGG or ResNet deep neural network structure.

[0015] Preferably, in the defect intelligent detection model, a non-maximum suppression method is used to find the best target bounding box to eliminate redundant bounding boxes.

[0016] Preferably, when identifying defects in the test sample detection image, the X-ray detection image is first segmented, then detected one by one, and finally the images marked with defects are spliced ​​into one image according to the original position for output.

[0017] The present invention also discloses a computer program product, comprising a computer program, wherein the computer program executes the steps of the method described above when executed by a processor.

[0018] The present invention further discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are executed.

[0019] The present invention also discloses an intelligent detection system for internal defects of power capacitors based on ray detection, comprising a memory and a processor connected to each other, wherein a computer program is stored on the memory, and when the computer program is run by the processor, the steps of the method described above are executed.

[0020] Compared with the prior art, the advantages of the present invention are:

[0021] The present invention adopts a deep neural network model to automatically identify internal defects of power capacitors from the X-ray detection images of the power capacitors, which can realize batch non-destructive detection of internal defects of power capacitors and realize the intelligent non-destructive detection process of power capacitors, while having high recognition accuracy and high recognition efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The figure is a flow chart of an embodiment of the intelligent detection method of the present invention.

[0023] Figure 2 Schematic diagram of sample images in the present invention; (a) is a diagram of a power capacitor; (b) is an upper transillumination image; and (c) is a lower transillumination image.

[0024] Figure 3 Schematic diagrams of typical defects of the power capacitor in the present invention; (a) is a solder joint porosity; (b) is internal shedding; (c) is an internal crack.

[0025] Figure 4 This is a schematic diagram of expanding a picture by random cropping in the present invention.

[0026] Figure 5 The deep neural network architecture diagram in the present invention; (a) is the SSD model architecture diagram, and (b) is the attention mechanism module (CBAM) architecture diagram.

[0027] Figure 6 Schematic diagram of the picture detection operation in the present invention. DETAILED DESCRIPTION

[0028] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0029] The intelligent detection method for internal defects of a power capacitor based on ray detection according to an embodiment of the present invention specifically comprises the following steps:

[0030] S1: Obtaining radiographic detection images: Using radiographic detection equipment (X-ray machine or industrial CT, etc.) with sufficient penetrating power to detect the sample, obtain transillumination images of different parts and angles and save them;

[0031] S2: Image preprocessing: preprocess the radiographic images obtained by radiographic detection, including image filtering and enhancement, to obtain clear and complete sample radiographic images;

[0032] S3: Image marking: Mark the sample radiographic images one by one, use boxes to mark the location and type of defects in the images, and finally generate a data set in VOC format;

[0033] S4: Dataset expansion: Expand the existing dataset by random cropping and free transformation, greatly expanding the existing sample size;

[0034] S5: training a deep convolutional neural network model: building an improved SSD (Single shot multibox detector) deep neural network model, and using the data set generated in step S4 to train the improved SSD model to obtain a trained model (defect intelligent detection model);

[0035] Specifically, the improved SSD deep neural network model consists of three parts: backbone network layer, additional feature layer and feature classification and regression layer. The backbone network layer is used to compress images and extract feature information from images; the additional feature layer is a network used to derive multi-scale feature maps; the feature classification and regression layer is used to classify and regress anchor boxes; an attention mechanism module is added after the backbone network layer to enhance the network's ability to identify small defects; finally, the non-maximum suppression method is used to find the best target bounding box and eliminate redundant bounding boxes. The backbone network adopts mainstream deep neural network structures such as MobileNet, VGG, and ResNet; the attention mechanism module adopts the CBAM (Convolutional Block Attention Module) module.

[0036] S6: Defect identification: Use the model trained in step S5 (defect intelligent detection model) to identify the type and number of defects in the test sample detection image.

[0037] The algorithm based on the DL model can not only automatically extract the features of the image and achieve high-precision detection and segmentation, but also has excellent generalization. Therefore, it is of great application value in radiographic image processing and is of great significance to improving the detection efficiency and objectivity of the detection results. Therefore, the present invention adopts the DL model to automatically identify and mark the defects of the power capacitor from the radiographic detection image, that is, to identify the defects inside the power capacitor through radiographic detection + SSD deep neural network model, which greatly improves the reliability of defect identification.

[0038] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0039] like Figure 1 As shown, the intelligent detection method for internal defects of power capacitors based on ray detection of the present invention has the following implementation steps:

[0040] S1. Select an X-ray machine or industrial CT machine of corresponding energy level according to the size and thickness of the power capacitor, such as 450kV industrial CT. Use the selected equipment to detect the power capacitor, obtain transillumination pictures of different parts and angles, and select pictures with clear images and complete structures for storage. In this embodiment, 20 power capacitors are selected for scanning, and two transillumination pictures are obtained for each, and a total of 40 sample pictures are obtained, such as Figure 2 shown.

[0041] S2. Use image processing techniques such as image filtering and enhancement to process the transillumination image to filter out image noise and make the internal defect features clearer.

[0042] S3. Labelme software is used to mark the enhanced sample images one by one, and boxes are used to mark the location and type of defects in the images. Finally, a VOC format data set is generated. Typical defect examples are shown in the figure below. Figure 3 shown.

[0043] S4. Based on the generated VOC format dataset, the image is expanded by random cropping. The implementation process is as follows: Figure 4 As shown in the figure, the original image is randomly cropped using a rectangular frame of 512 pixels × 512 pixels, and 200 new images are generated for each original image (4096 pixels × 4096 pixels). The invalid images that do not contain defects are removed using an algorithm, and then a new VOC format dataset is generated on this basis. The dataset at this time contains more than 3,000 sample images, and the training set, validation set, and test set are divided according to the ratio of 8:1:1.

[0044] S5. Build an SSD (Single shot multibox detector) deep neural network model and use the generated data set to train the SSD model. The structure of the SSD deep neural network is as follows: Figure 5 As shown in (a), it includes three parts: a backbone network layer, an additional feature layer, and a feature classification and regression layer. The backbone network layer is used to compress images and extract feature information from images. The general ResNet50 neural network structure is selected here; the additional feature layer is a network used to derive multi-scale feature maps; the feature classification and regression layer is used to perform target classification and position regression on anchor boxes; finally, the non-maximum suppression method is used to find the best target bounding box and eliminate redundant bounding boxes. It should be noted that in order to enhance the network's ability to recognize small defects, the present invention adds an attention mechanism module (CBAM) after the backbone network layer. CBAM (Convolutional Block Attention Module) is an attention mechanism used for convolutional neural networks (CNNs), and its structure is as shown in the figure. Figure 5As shown in (b), CBAM mainly consists of two sub-modules: Channel Attention Module and Spatial Attention Module. CBAM can enhance the network's attention to input features and improve the network's ability to identify small defects.

[0045] S6. Use the trained SSD model to predict defects in the test sample image. Since the input image size is large and the defects therein are small and dense, the present invention first divides the test image (4096 pixels × 4096 pixels) into 64 512 pixels × 512 pixels images in 8 × 8 mode, then detects them one by one, and finally splices the 64 defect-marked images into one image according to the original position for output. The operation process is as follows: Figure 6 By dividing the large-size image into small-size images and then performing defect detection on the small images, the operation speed is improved while suppressing the noise in most areas of the image.

[0046] The present invention adopts a deep neural network model to automatically identify internal defects of power capacitors from their X-ray detection images, thereby solving the difficulties in batch nondestructive detection of internal defects of power capacitors. At the same time, it solves the problems of low efficiency, strong subjectivity and easy misjudgment of manual determination of defects, thereby realizing an intelligent nondestructive detection process for power capacitors.

[0047] The present invention constructs an intelligent nondestructive testing method for power capacitors based on X-ray nondestructive testing, and realizes the process of intelligent nondestructive testing of internal defects of power capacitors; the method of the present invention can not only greatly improve the efficiency of testing and greatly reduce labor costs, but also make the test results more objective and reliable.

[0048] The present invention also discloses a computer program product, including a computer program, which executes the steps of the above method when executed by a processor. The present invention further discloses a computer-readable storage medium, on which a computer program is stored, which executes the steps of the above method when executed by a processor. The present invention also discloses an intelligent detection system for internal defects of power capacitors based on X-ray detection, including an interconnected memory and a processor, on which a computer program is stored, which executes the steps of the above method when executed by a processor. The product, medium and system of the present invention correspond to the above method and also have the advantages described in the above method.

[0049] The present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by hardware related to computer program instructions. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned method embodiment when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable storage media include: any entity or device that can carry computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. The memory is used to store computer programs and / or modules, and the processor implements various functions by running or executing computer programs and / or modules stored in the memory, and calling data stored in the memory. The memory may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, an internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0050] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. An intelligent detection method for internal defects of power capacitors based on X-ray detection, characterized in that: Includes steps: Obtaining radiographic inspection images of different parts and angles of the power capacitor and preprocessing the radiographic inspection images; Mark the preprocessed X-ray detection images, mark the locations and types of defects in the X-ray detection images, and finally generate a data set; Build a deep neural network model and use the data set to train the deep neural network model to obtain a trained defect intelligent detection model; The trained defect intelligent detection model is used to identify the type and quantity of defects in the test sample inspection image.

2. The method for intelligent detection of internal defects of power capacitors based on ray detection according to claim 1 is characterized in that: The X-ray detection images are processed by using image filtering and enhanced image processing techniques to filter out image noise.

3. The intelligent detection method for internal defects of power capacitors based on ray detection according to claim 1 is characterized in that: After generating the dataset, it is expanded by random cropping and free transformation.

4. The intelligent detection method for internal defects of power capacitors based on ray detection according to claim 1, 2 or 3, characterized in that: The deep neural network model includes three parts: a backbone network layer, an additional feature layer, and a feature classification and regression layer; wherein the backbone network layer is used to compress images and extract feature information from images; The additional feature layer is used to introduce a network of multi-scale feature maps; the feature classification and regression layer is used to perform target classification and position regression on the anchor box; an attention mechanism module is added after the backbone network layer to enhance the network's ability to recognize small defects.

5. The method for intelligent detection of internal defects of power capacitors based on ray detection according to claim 4 is characterized in that: The backbone network layer adopts a MobileNet, VGG or ResNet deep neural network structure.

6. The method for intelligent detection of internal defects of power capacitors based on ray detection according to claim 4 is characterized in that: In the defect intelligent detection model, the non-maximum suppression method is used to find the best target bounding box to eliminate redundant bounding boxes.

7. The method for intelligent detection of internal defects of power capacitors based on ray detection according to claim 1, 2 or 3, characterized in that: When identifying defects in the test sample inspection image, the X-ray inspection image is first segmented, then inspected one by one, and finally the images marked with defects are spliced ​​into one image according to the original position for output.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are performed.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program performs the steps of the method according to any one of claims 1 to 7.

10. An intelligent detection system for internal defects of power capacitors based on X-ray detection, comprising a memory and a processor connected to each other, wherein a computer program is stored in the memory, characterized in that: When the computer program is executed by a processor, the computer program performs the steps of the method according to any one of claims 1 to 7.

Images