Epoxy resin defect identification method and system based on improved convolutional neural network

Abstract

The invention relates to the technical field of power equipment partial discharge detection, in particular to an epoxy resin defect identification method and system based on an improved convolutional neural network. The method comprises the following steps: carrying out a partial discharge test on an epoxy resin defect sample, collecting a partial discharge signal, carrying out primary transformation on the partial discharge signal to obtain a partial discharge signal three-dimensional time-frequency characteristic pattern, and identifying the time-frequency characteristic pattern by adopting a first algorithm to realize identification of the epoxy resin defect type. According to the method, the attention mechanism structural block is inserted into the CNN, so that the attention of the algorithm on high-frequency plaques in a time-frequency graph and unique time-frequency spectrum features of different discharge fault types is effectively improved, the model can pay more attention to key features in the image, the generalization ability of the model on new data is improved, and the accuracy of the model is improved. And the epoxy resin defect type identification accuracy of the model is improved.

Description

Technical Field

[0001] The invention belongs to the field of partial discharge detection of power equipment, and specifically relates to an epoxy resin defect recognition method based on an improved convolutional neural network. Background Art

[0002] Epoxy resin potting materials have excellent mechanical properties, electrical insulation properties and stable chemical properties. At the same time, due to its simple preparation process and easy molding, it has been widely used in the fields of electronic circuit potting, device insulation and moisture-proofing. In the process of epoxy resin pouring and curing, even with vacuum degassing, a certain number of bubbles with heat insulation effect are still inevitable, and as the volume fraction of filler increases, the interference of gas is more obvious. In addition to affecting the appearance, the presence of bubbles also seriously affects the reliability of potting parts. Since the dielectric constant of gas is lower than that of the surrounding solid insulating material, and the breakdown field strength of gas is usually lower than that of solid insulating material, gas breakdown will first occur in the air gap under the action of applied voltage, but the surrounding solid medium still maintains insulation properties. The discharge in these defects may develop into tree discharge, which eventually leads to insulation breakdown. Therefore, from a practical and technical point of view, it is very important to detect and identify internal defects in epoxy resin.

[0003] Although some studies have focused on the partial discharge phenomenon of epoxy resin, relatively few studies have been conducted on the identification of different types of epoxy resin defects. This is mainly because the defects of epoxy resin are diverse in morphology and difficult to observe directly inside the material. The application of machine vision inspection in the detection of epoxy resin defect types has a series of advantages, but also faces some challenges. Machine vision systems can accurately inspect products, avoid human errors, and improve product quality. It can measure tiny sizes with an accuracy of one thousandth of an inch. However, for some specific types of defects, such as tiny bubbles or subsurface defects, machine vision systems may have difficulty detecting them. In addition, existing image processing and pattern recognition algorithms may not meet the needs of all types of defect detection, especially when the defect features are not obvious or the background is complex. With the development of artificial intelligence technology, deep learning-based methods have been widely used in product defect detection due to their excellent feature extraction capabilities. Methods such as deep convolutional neural networks can learn high-order features from raw data to achieve high recognition accuracy. Although convolutional neural networks (CNNs) have achieved remarkable results in defect detection, they still have some shortcomings and challenges. For example, CNN models usually require a large amount of labeled data for training, and the quality and diversity of the data directly affect the performance and generalization ability of the model. For defect detection, it may be difficult to obtain enough high-quality defect samples.

[0004] Based on the above analysis, this paper proposes an epoxy resin defect recognition method based on an improved convolutional neural network. The three-dimensional time-frequency graph of partial discharge of different defects is used as data input, and the attention mechanism structure block is inserted into the CNN to effectively improve the algorithm's attention to the high-frequency patches in the time-frequency graph and the unique time-frequency characteristics of different discharge fault types, so that the model can pay more attention to the key features in the image and improve the model's generalization ability to new data. By accurately identifying tiny defects in epoxy resin, this method can take timely measures to avoid potential partial discharge and insulation failure, thereby significantly improving the reliability and safety of the power system, which is crucial for preventing equipment failures and ensuring the stable operation of the power grid. Summary of the invention

[0005] In view of the above problems existing in the prior art, the present invention is proposed.

[0006] To solve the above technical problems, the present invention provides the following technical solution, which is an epoxy resin defect recognition method based on an improved convolutional neural network, including: performing a partial discharge test on an epoxy resin defect sample, collecting a partial discharge signal, performing a primary transformation on the partial discharge signal to obtain a three-dimensional time-frequency characteristic diagram of the partial discharge signal, and using a first algorithm to identify the time-frequency characteristic diagram to realize the recognition of the epoxy resin defect type.

[0007] As a preferred solution of the epoxy resin defect recognition method based on the improved convolutional neural network described in the present invention, wherein: the partial discharge test on the epoxy resin defect sample includes:

[0008] The partial discharge test is carried out under repetitive unipolar square wave excitation.

[0009] As a preferred solution of the epoxy resin defect recognition method based on improved convolutional neural network described in the present invention, wherein: the collecting of partial discharge signals includes using a UHF antenna to collect partial discharge signals.

[0010] As a preferred solution of the epoxy resin defect recognition method based on improved convolutional neural network described in the present invention, wherein: obtaining the three-dimensional time-frequency characteristic diagram of the partial discharge signal includes drawing the three-dimensional time-frequency characteristic diagram of the partial discharge using a function of related software.

[0011] As a preferred solution of the epoxy resin defect recognition method based on the improved convolutional neural network described in the present invention, the first-level transformation of the partial discharge signal includes using a function in the relevant software to perform a first-level transformation on the partial discharge signal.

[0012] As a preferred solution of the epoxy resin defect identification method based on the improved convolutional neural network described in the present invention, the partial discharge test is carried out under the excitation of a repetitive unipolar square wave, including that the rise and fall time of the repetitive unipolar square wave excitation is 1 μs, the frequency is 500 Hz, and the duty cycle is 50%.

[0013] As a preferred solution of the epoxy resin defect recognition method based on improved convolutional neural network described in the present invention, wherein: the UHF antenna range is 50kHz~6.5GHz.

[0014] As a preferred solution of the epoxy resin defect recognition system based on the improved convolutional neural network described in the present invention, it includes: a partial discharge acquisition module, a time-frequency feature generation module, and a defect type recognition module.

[0015] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of any one of the methods for epoxy resin defect recognition based on an improved convolutional neural network are implemented.

[0016] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of any one of the methods for epoxy resin defect recognition based on an improved convolutional neural network are implemented.

[0017] The beneficial effects of the present invention are as follows: (1) The present invention uses repetitive square wave excitation to perform partial discharge tests. Repetitive square wave excitation combines the advantages of industrial frequency AC excitation and impulse excitation, which can better stimulate defects to produce discharges and help collect more key features of partial discharges of defects. In addition, the original discharge data can be plotted into a three-dimensional time-frequency graph by using MATLAB functions such as scalogram and pcolor. Using a three-dimensional time-frequency graph as data input for CNN has many advantages. The image contains rich visual information, such as color, texture, shape, etc., which is crucial for image recognition and classification tasks. In addition, CNN can automatically extract features from images without manually designing feature extraction algorithms, simplifying the image processing process.

[0018] (2) The present invention inserts an attention mechanism structural block into CNN, which effectively improves the algorithm's attention to the high-frequency patches in the time-frequency diagram and the unique time-frequency spectrum features of different discharge fault types, so that the model can pay more attention to the key features in the image, improve the model's generalization ability for new data, and thus improve the model's accuracy in identifying epoxy resin defect types. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0020] Figure 1 A schematic diagram of epoxy resin bubble, crack and surface delamination defect samples according to an epoxy resin defect recognition method based on an improved convolutional neural network provided in one embodiment of the present invention.

[0021] Figure 2 A schematic diagram of a partial discharge test platform for an epoxy resin defect recognition method based on an improved convolutional neural network provided in one embodiment of the present invention.

[0022] Figure 3 A schematic diagram of the time domain, frequency domain and three-dimensional time-frequency analysis results of bubble defects in an epoxy resin defect recognition method based on an improved convolutional neural network provided by an embodiment of the present invention.

[0023] Figure 4 A schematic diagram of the time domain, frequency domain and three-dimensional time-frequency analysis results of crack defects in an epoxy resin defect recognition method based on an improved convolutional neural network provided by an embodiment of the present invention.

[0024] Figure 5 A schematic diagram of the time domain, frequency domain and three-dimensional time-frequency analysis results of surface delamination defects of an epoxy resin defect recognition method based on an improved convolutional neural network provided in one embodiment of the present invention.

[0025] Figure 6 A schematic diagram of the function of an attention mechanism structural block (CBAM) of an epoxy resin defect recognition method based on an improved convolutional neural network provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0027] Example 1

[0028] Reference Figure 1-Figure 6 , which is the first embodiment of the present invention, and provides an epoxy resin defect recognition method based on an improved convolutional neural network, such as Figure 1 As shown, including:

[0029] It should be noted that a partial discharge test is performed on an epoxy resin defective sample, a partial discharge signal is collected, a primary transformation is performed on the partial discharge signal to obtain a three-dimensional time-frequency characteristic diagram of the partial discharge signal, and the first algorithm is used to identify the time-frequency characteristic diagram to realize the identification of the epoxy resin defect type.

[0030] In the embodiment of the present application, the first-level transformation uses Fourier transform to obtain a three-dimensional time-frequency characteristic diagram of the partial discharge signal.

[0031] Most preferably, the CER6521 epoxy resin impregnation system is used to prepare Figure 1 The epoxy resin bubbles, cracks and surface delamination defects shown in the figure are set to 500Hz, the rise time is 1μs, and the duty cycle is 50%. During the experiment, a UHF antenna is used to collect partial discharge signals. Figure 2 The partial discharge test platform shown in the figure. In order to avoid surface discharge during the experiment, the defective sample and the electrode are immersed in the oil cup as a whole. The partial discharge test is carried out according to the standard test process, and 100 sets of partial discharge signals are recorded for each defect. After the test is completed, the three-dimensional time-frequency diagram of the partial discharge signal under different defects is drawn using the fft, scalogram and pcolor functions in the MATLAB software. Figure 3 The time domain, frequency domain and three-dimensional time-frequency analysis results of bubble defects. Figure 4 The time domain, frequency domain and three-dimensional time-frequency analysis results of crack defects. Figure 5 The time domain, frequency domain and three-dimensional time-frequency analysis results of surface delamination defects. From the three-dimensional time-frequency diagram, it can be seen that the partial discharge signals of the three defects have obvious differences, and this difference feature can be used to identify the defect type.

[0032] Furthermore, in order to obtain more data input, a batch generation algorithm for time-frequency graphs is used to construct a sample library of local discharge three-dimensional time-frequency graphs. The number of time-frequency graphs for each defect is 2,000, 80% of the time-frequency graphs are selected as training sets, and 20% of the time-frequency graphs are selected as test sets. Although convolutional neural networks (CNNs) have achieved remarkable results in defect detection, there are still some shortcomings and challenges. This paper proposes an epoxy resin defect recognition method based on an improved convolutional neural network. The three-dimensional time-frequency graphs of local discharges of different defects are used as data input, and the attention mechanism structure block is inserted into CNN. It effectively improves the algorithm's attention to the high-frequency patches in the time-frequency graph and the unique time-frequency characteristics of different discharge fault types, so that the model can pay more attention to the key features in the image and improve the model's generalization ability to new data.

[0033] In an optional embodiment, the first-level transformation may be: Wavelet transform is a tool suitable for non-stationary signal analysis, which can provide local features of signals at different time and frequency scales, thereby generating a three-dimensional time-frequency feature map. Its application process is:

[0034] Furthermore, wavelet transform can analyze the frequency components of signals at multiple scales, which is particularly suitable for capturing the transient characteristics of epoxy resin defect signals. Unlike Fourier transform, wavelet transform can extract specific frequency features in a short time.

[0035] Furthermore, since the local discharge characteristics of different defects are distributed differently in time and frequency, wavelet transform can better represent the characteristics of defect signals by separating high- and low-frequency features, providing high-quality time-frequency feature input for subsequent defect classification.

[0036] Furthermore, the three-dimensional time-frequency diagram generated by wavelet transform retains the changing characteristics of different defects in the time domain and frequency domain, and can clearly present the unique signal characteristics of defects such as bubbles, cracks and surface delamination in the image, thus providing effective data support for defect type identification.

[0037] In an optional embodiment, the first-level transformation can be that the short-time Fourier transform obtains the frequency distribution of the signal at different times by performing the Fourier transform in a fixed window, thereby generating a three-dimensional time-frequency feature map. Although STFT has a certain time and frequency resolution, it has limitations in extracting time-frequency features due to the fixed window length.

[0038] It should be noted that the partial discharge signal is divided into small segments, and the spectrum characteristics of each segment are analyzed using Fourier transform. By merging the transformation results of all small segments, a three-dimensional time-frequency diagram can be generated.

[0039] It should be noted that compared with Fourier transform, the time-frequency diagram generated by STFT is weaker in expressing time-varying characteristics, and the model recognition accuracy may be slightly lower, but it can still distinguish epoxy resin defect types to a certain extent.

[0040] In an embodiment of the present application, the first algorithm uses an improved convolutional neural network to identify the time-frequency feature map, thereby realizing the identification of epoxy resin defect types.

[0041] It should be noted that in the embodiments of the present application, the architecture of the convolutional neural network (CNN) used is a residual network (ResNet).

[0042] Furthermore, in the method of the present invention: a batch generation algorithm of time-frequency graphs is used to construct a sample library of partial discharge three-dimensional time-frequency graphs. The number of time-frequency graphs for each defect is 2000, 80% of the time-frequency graphs are selected as training sets, and 20% of the time-frequency graphs are selected as test sets.

[0043] Furthermore, in the method of the present invention: improving the convolutional neural network is to insert an attention mechanism structural block (CBAM) into the convolutional neural network.

[0044] Furthermore, in the method of the present invention: the attention mechanism is a resource allocation strategy that allows the model to focus on the most important parts when processing information. Inserting CBAM in CNN can enhance the network's ability to focus on important features in the input data.

[0045] Furthermore, the architecture of the convolutional neural network (CNN) used is the residual network (ResNet). ResNet, by introducing a residual learning framework, allows the gradient to flow directly to the shallower network, effectively alleviating the gradient vanishing problem that occurs as the network deepens. Due to the existence of residual connections, ResNet can easily increase the depth of the network without compromising the training performance, which makes it possible to build a deeper network. In addition, the structure of ResNet is relatively simple, mainly consisting of convolutional layers, batch normalization layers, and ReLU activation functions, which makes the network easier to implement and understand. For residual networks, due to the introduction of residual connections, the network can learn the identity mapping at a shallower level, which can speed up convergence and improve training efficiency. The attention mechanism can make the network focus on the most important areas or features in the image, thereby improving the expressiveness of the features and helping to extract richer and more discriminative features. By focusing on the key parts of the input data, the attention mechanism helps to improve the performance of the model in tasks such as image classification, object detection, and image segmentation. In addition, the attention mechanism allows the model to learn different feature attention patterns on different input samples, which may improve the model's generalization ability to new data.

[0046] Figure 6The figure is a schematic diagram of the attention mechanism building block (CBAM). As can be seen from the figure, CBAM can identify the most informative feature channels by applying the attention mechanism on the channel dimension of the time-frequency graph, which helps the model to focus on learning the frequency and time features that are critical to the task. In the two-dimensional spatial direction, CBAM assigns weights to each spatial position in the time-frequency graph, allowing the model to accurately locate the key areas in the signal. The adaptive weight allocation of CBAM reduces the processing of unimportant information, thereby accelerating the training process of the model and enabling the network to learn the key information in the time-frequency graph more quickly. By strengthening the features of key channels and spatial positions, CBAM significantly improves the recognition accuracy of the model for the time-frequency graph and greatly improves the accuracy of defect type recognition. Through comparative tests, it can be seen that the accuracy of defect type recognition before the introduction of CBAM is about 96.4%, while the accuracy of defect type recognition after the introduction of CBAM is about 99.2%. It can be seen that this method can effectively realize the recognition of internal defect types of epoxy resin.

[0047] In an optional embodiment, the first algorithm may be a support vector machine (SVM), which is a classic classification algorithm that has good classification effect when the features are clear and the amount of data is relatively small. This algorithm mainly relies on manually extracted signal features for identification and classification.

[0048] Manually extract several features from the three-dimensional time-frequency graph, including frequency domain features (such as frequency peak, mean frequency), time domain features (such as peak, standard deviation), and time-frequency joint features. Use feature selection algorithms (such as principal component analysis, PCA) to reduce the dimension of manual features to reduce redundant information and improve classification accuracy. Establish an SVM classifier based on the selected features, using 80% of the time-frequency graph feature data as a training set and 20% as a test set. After manual feature extraction and feature selection, the new data is input into the SVM model for defect type identification.

[0049] It should be noted that although SVM has strong classification ability in feature space division, its effect depends on the effectiveness of manual features. Manual features are not as effective as improved CNN automatic learning features in extracting defect details.

[0050] In an optional embodiment, the first algorithm may be to use a standard two-dimensional convolutional neural network (2D-CNN) to analyze and classify the three-dimensional time-frequency graph of epoxy resin defect types. Compared with the improved ResNet network, the standard 2D-CNN model does not have a residual connection structure and does not contain an attention mechanism module (CBAM), so it is relatively weak in recognition performance and detail processing.

[0051] The collected three-dimensional time-frequency feature map is resized to generate a two-dimensional image input of uniform size. A standard 2D-CNN network structure is constructed, including convolutional layers, pooling layers, and fully connected layers, for feature extraction and classification of the time-frequency map. 2D-CNN extracts the spatial features of the image through simple convolution and pooling operations. 80% of the time-frequency map data is used as a training set and 20% as a test set. During the training process, CNN automatically learns different features of the time-frequency map to distinguish different defect types. The test data is input into the trained CNN model and the predicted defect type is output.

[0052] It should be noted that although the standard 2D-CNN can recognize defect types, it is weak in capturing high-frequency or local features in the time-frequency graph, especially in the case of defects with small details or obvious feature variations, and its recognition ability is not as good as the improved ResNet plus CBAM network structure. Therefore, the recognition accuracy and generalization ability of this method for new data are lower than those of the improved convolutional neural network solution.

[0053] Example 2

[0054] The second embodiment of the present invention is different from the previous embodiment in that:

[0055] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0056] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0057] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0058] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0059] Example 3

[0060] The third embodiment of the present invention provides an epoxy resin defect recognition system based on an improved convolutional neural network, which is characterized by including a local discharge acquisition module, a time-frequency feature generation module, and a defect type recognition module.

[0061] Partial discharge acquisition module, to conduct partial discharge test on epoxy resin defective samples;

[0062] The time-frequency feature generation module collects the partial discharge signal and performs a first-level transformation on the partial discharge signal to obtain a three-dimensional time-frequency feature diagram of the partial discharge signal;

[0063] The defect type identification module uses the first algorithm to identify the time-frequency characteristic diagram to realize the identification of the defect type of the epoxy resin.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. The epoxy resin defect recognition method based on improved convolutional neural network is characterized by: include, A partial discharge test is performed on the epoxy resin defective sample, and the partial discharge signal is collected. The partial discharge signal is transformed into a primary state to obtain a three-dimensional time-frequency characteristic diagram of the partial discharge signal. The first algorithm is used to identify the time-frequency characteristic diagram to realize the identification of the epoxy resin defect type.

2. The epoxy resin defect recognition method based on the improved convolutional neural network according to claim 1, characterized in that: The partial discharge test on the epoxy resin defective sample includes: The partial discharge test is carried out under repetitive unipolar square wave excitation.

3. The epoxy resin defect recognition method based on improved convolutional neural network according to claim 2, characterized in that: The collecting of the partial discharge signal includes collecting the partial discharge signal using a UHF antenna.

4. The epoxy resin defect recognition method based on improved convolutional neural network according to claim 3, characterized in that: The obtaining of the three-dimensional time-frequency characteristic diagram of the partial discharge signal includes drawing the three-dimensional time-frequency characteristic diagram of the partial discharge signal using a function of relevant software.

5. The epoxy resin defect recognition method based on improved convolutional neural network according to claim 4, characterized in that: The performing a primary transformation on the partial discharge signal includes performing a primary transformation on the partial discharge signal using a function in relevant software.

6. The epoxy resin defect recognition method based on improved convolutional neural network according to claim 5, characterized in that: The partial discharge test is carried out under repetitive unipolar square wave excitation, wherein the rise and fall time of the repetitive unipolar square wave excitation is 1 μs, the frequency is 500 Hz, and the duty cycle is 50%.

7. The epoxy resin defect recognition method based on improved convolutional neural network according to claim 6, characterized in that: The UHF antenna has a range of 50kHz to 6.5GHz.

8. A system for epoxy resin defect recognition method based on improved convolutional neural network according to any one of claims 1 to 7, characterized in that: It includes a local discharge acquisition module, a time-frequency feature generation module, and a defect type identification module.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, 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 implemented.

Images