A Defect Detection Method for Transmission Hardware Based on Cascade Network

Abstract

The present invention discloses a method for detecting defects of transmission small fittings based on a cascaded network, belonging to the technical field of defect detection of transmission small fittings. The specific method includes: Step 1: Obtain the production line assembly drawings of transmission small fittings; set up the detection device for transmission small fittings; Step 2: Obtain several representative images of transmission small fittings, perform representative image processing, and obtain several groups of training sets and test sets; Step 3: Construct a convolutional neural network for detecting defects of transmission small fittings, and then train and test the convolutional neural network through the obtained training sets and test sets, and mark the successfully tested convolutional neural network as a defect detection model; Step 4: Obtain the detection data of the working simulation module in the detection device, judge whether the detection data is qualified. When the detection data is unqualified, record the corresponding unqualified area; when the detection data is qualified, proceed to the next step; Step 5: Detect the transmission small fittings passing through the detection area through the defect detection model.

Description

Technical Field

[0001] The present invention belongs to the technical field of defect detection of transmission hardware, and specifically relates to a method for defect detection of transmission hardware based on a cascade network. Background Art

[0002] The power industry is a pillar industry for economic development. The safe and stable operation of the power system has great strategic significance. Among them, defects in the components of transmission hardware play an important role in the safety of transmission lines. Therefore, how to timely detect defects in the production process of transmission hardware and avoid the potential safety hazards and corresponding difficulties in replacement caused by installing defective transmission hardware into the power grid.

[0003] Therefore, there is currently a need for a method for defect detection of transmission hardware based on a cascade network to achieve timely detection of defects in transmission hardware during the production process. Summary of the Invention

[0004] In order to solve the problems existing in the above solutions, the present invention provides a method for defect detection of transmission hardware based on a cascade network.

[0005] The object of the present invention can be achieved by the following technical solutions:

[0006] A method for defect detection of transmission hardware based on a cascade network, the specific method includes:

[0007] Step 1: Obtain the assembly drawings of the production line of the transmission hardware; set up the detection device for the transmission hardware.

[0008] Step 2: Obtain several representative images of the transmission hardware, perform representative image processing, and obtain several groups of training sets and test sets.

[0009] Step 3: Construct a convolutional neural network for defect detection of transmission hardware, and then train and test the convolutional neural network through the obtained training sets and test sets, and mark the successfully tested convolutional neural network as a defect detection model.

[0010] Step 4: Obtain the detection data of the working simulation module in the detection device, determine whether the detection data is qualified. When the detection data is unqualified, record the corresponding unqualified area; when the detection data is qualified, proceed to the next step.

[0011] Step 5: Detect the transmission hardware in the detected area through the defect detection model. When the detection result is qualified, no operation is performed; when the detection result is unqualified, record the corresponding unqualified area.

[0012] Step 6: Sort the unqualified transmission hardware.

[0013] Further, the method for setting up the detection device for transmission line hardware includes:

[0014] Identify the assembly drawings of the transmission line hardware production line, divide the detection area, set up a housing device in the detection area, with multi-color light sources installed inside the housing device. Identify the types, models, and working environments of the transmission line hardware that need to be detected. Set up a working simulation module inside the housing device according to the working environment of the transmission line hardware. Set up an image acquisition module inside the housing device, obtain the current detection environment, assign the obtained types, models, working environments of the transmission line hardware, and the current detection environment, and convert them into matching vectors. Establish a defect detection solution library, input the matching vectors into the defect detection solution library, match the corresponding defect detection solutions, and supplement the detection device according to the matched defect detection solutions.

[0015] Further, the working method of the working simulation module includes:

[0016] Obtain the working environment of the transmission line hardware, set up a corresponding voltage simulation environment according to the obtained working environment, and establish a corresponding simulation matching table; identify the working environment of the current transmission line hardware, match the corresponding voltage simulation environment from the simulation matching table, detect the current transmission line hardware according to the matched voltage simulation environment, and generate an image acquisition signal during the detection process and send it to the image acquisition module.

[0017] Further, the working method of the image acquisition module includes:

[0018] When receiving the image acquisition signal, control the multi-color light sources inside the housing device to turn on the corresponding color light sources; perform image acquisition, mark it as the initial acquisition image, and perform background segmentation on the initial acquisition image to obtain the corresponding acquisition image of the transmission line hardware.

[0019] Further, the method for establishing the defect detection solution library includes:

[0020] Obtain the target information of the transmission line hardware, where the target information includes the types, models, and working environments of the transmission line hardware. Classify the obtained target information to obtain classification data. Develop a defect detection solution based on the housing device according to the classification data. Extract the characteristic vectors of the defect detection solution, establish a first database, store the set defect detection solution in the first database, establish a vector space unit in the first data, map the extracted characteristic vectors to the vector space unit, and mark the current first database as the defect detection solution library.

[0021] Further, the method for performing representative image processing includes:

[0022] Perform super-resolution on the representative image based on the image super-resolution algorithm, mark the super-resolved representative image as the super-resolved image, mark the qualified areas and defective areas in the super-resolved image, and mark the corresponding position information on the defective areas. Segment the super-resolved image based on a preset size to obtain several segmented images, identify the marked information on the segmented images, and integrate them into a training set and a test set.

[0023] Further, the method for sorting out unqualified transmission hardware fittings includes:

[0024] Construct a transmission hardware fitting model, divide different sorting areas in the transmission hardware fitting model, and set corresponding transmission paths and storage locations for each sorting area; obtain the unqualified areas corresponding to the transmission hardware fittings with defects, match the corresponding sorting areas, and transfer the transmission hardware fittings with defects to the storage locations through the corresponding transmission paths according to the matched sorting areas.

[0025] Further, the housing device is used to block light and isolate the outside world.

[0026] Compared with the prior art, the beneficial effects of the present invention are:

[0027] By setting the housing device, it is avoided that during the detection process, the production operation and safety of the outside world are affected due to the need to simulate the working environment; and through the mutual cooperation of each detection device, the detection of transmission hardware fittings in the simulated working environment is realized, making the detection more comprehensive and specific; by setting a multi-color light source, more accurate background segmentation is achieved; the unqualified transmission hardware fittings are sorted out, which is convenient for subsequent separate processing of the defective transmission hardware fittings, improving production efficiency and resource utilization rate. Description of the Drawings

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is the flowchart of the method of the present invention. Detailed Embodiments

[0030] The following will clearly and completely describe the technical solutions of the present invention in combination with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0031] As Figure 1 shown, a defect detection method for transmission line hardware based on a cascaded network, the specific method includes:

[0032] Step 1: Obtain the production line assembly drawing of the transmission line hardware; set up the detection device for the transmission line hardware based on the obtained production line assembly drawing of the transmission line hardware.

[0033] The method for setting up the detection device for the transmission line hardware includes:

[0034] Identify the production line assembly drawing of the transmission line hardware, divide the detection area, which can be directly divided manually according to production requirements, more efficiently; set up a housing device in the detection area, that is, the production line of the transmission line hardware passes through the housing device, and a multi-color light source is arranged inside the housing device, and the multi-color light source is used to emit various single-color lights as needed; identify the type, model and working environment of the transmission line hardware to be detected, set up a working simulation module inside the housing device according to the working environment of the transmission line hardware, and set up an image acquisition module inside the housing device, and the image acquisition module is used to acquire images according to the received signal; obtain the current detection environment, assign the obtained type, model, working environment of the transmission line hardware and the current detection environment and convert them into matching vectors, establish a defect detection solution library, input the matching vectors into the defect detection solution library, match the corresponding defect detection solution, and supplement the detection device according to the matched defect detection solution to complete the setting of the detection device for the transmission line hardware.

[0035] When the detection work needs to be carried out, data collection and detection are carried out according to the defect detection solution.

[0036] The housing device is used to block light and isolate the outside world, so as to form a closed environment at the internal detection position, and the specific structure can directly use existing devices and structures that can realize its function.

[0037] By setting up the housing device, it is avoided that during the detection process, the production operation and safety of the outside world are affected due to the need to simulate the working environment; and through the mutual cooperation of each detection device, the detection of the transmission line hardware in the simulated working environment is realized, making the detection more comprehensive and specific.

[0038] The working method of the working simulation module includes:

[0039] Obtain the working environment of the transmission small hardware fittings, set the corresponding voltage simulation environment according to the obtained working environment, which can be set by discussion by the expert group; establish a corresponding simulation matching table; identify the working environment of the current transmission small hardware fittings, match the corresponding voltage simulation environment from the simulation matching table, detect the current transmission small hardware fittings according to the matched voltage simulation environment, and generate an image acquisition signal during the detection process and send it to the image acquisition module.

[0040] The working method of the image acquisition module includes:

[0041] When receiving the image acquisition signal, control the multi-color light source in the housing device to turn on the corresponding color light source; perform image acquisition, mark it as the initial acquisition image, and perform background segmentation on the initial acquisition image to obtain the corresponding acquisition image of the transmission small hardware fittings.

[0042] The image acquisition module uses an image acquisition device that can work normally in the simulation environment.

[0043] Performing background segmentation on the initial acquisition image is a commonly used segmentation algorithm in this field, so it will not be described in detail. However, by setting the multi-color light source, more accurate background segmentation can be achieved.

[0044] The method for establishing a defect detection solution library includes:

[0045] Obtain the target information of the transmission small hardware fittings. The target information includes the type, model, and working environment of the transmission small hardware fittings. The working environment refers to the environment in which it works, such as the voltage at which it works; classify the obtained target information to obtain classified data, formulate a defect detection solution based on the housing device according to the classified data, extract the characteristic vector of the defect detection solution, establish a first database, store the set defect detection solution in the first database, establish a vector space unit in the first data, map the extracted characteristic vector into the vector space unit, and mark the current first database as the defect detection solution library.

[0046] Formulating a defect detection solution based on the housing device according to the classified data is set by the expert group according to different classified data and corresponding detection environments. The detection environment refers to the production and installation environment within the detection area; the defect detection solution is based on the summary statistics of the detection situations, and then relevant experts compile the corresponding defect detection solution according to the summary of the detection situations. The specific compilation process is common knowledge in this field, so it will not be described in detail.

[0047] The vector space unit is the vector space, and the corresponding characteristic vector can be matched according to the input vector.

[0048] The characteristic vector refers to the spatial vector obtained by assignment according to the target information in the defect detection scheme and the corresponding detection environment. The method of assignment is that the expert group pre-sets the assignment matching table of the corresponding target information and detection environment in advance, and then obtains the corresponding assignment through corresponding data matching, and then converts it into a characteristic vector.

[0049] The classification of the obtained target information is based on whether the corresponding transmission hardware fittings can be detected by the same defect detection scheme, and can be classified manually or by artificial intelligence.

[0050] Step 2: Obtain several representative images of the transmission hardware fittings. The representative images refer to the processed images that only contain the transmission hardware fittings and are the same as the images collected during the detection process; perform representative image processing to obtain several groups of training sets and test sets;

[0051] The methods for performing representative image processing include:

[0052] Perform super-resolution on the representative image based on the image super-resolution algorithm, mark the super-resolved representative image as the super-resolution image, mark the qualified area and the defect area in the super-resolution image, and mark the corresponding position information on the defect area, that is, mark which position on the transmission hardware fitting the defect area is located; segment the super-resolution image based on the preset size to obtain several segmented images. The preset size refers to the size set by the expert group according to the requirements of subsequent training of the convolutional neural network; identify the marked information on the segmented images and integrate them into the training set and the test set.

[0053] Step 3: Construct a convolutional neural network for defect detection of transmission hardware fittings, and then train and test the convolutional neural network through the obtained training set and test set, and mark the successfully tested convolutional neural network as the defect detection model; specifically, how to construct the convolutional neural network, train and test is common knowledge in this field, so it will not be described in detail; the defect detection model is used to judge whether the corresponding transmission hardware fitting has defects according to the input collected image, and output the position of the defect area when defects occur.

[0054] Step 4: Obtain the detection data of the working simulation module in the detection device, judge whether the detection data is qualified. When the detection data is unqualified, record the corresponding unqualified area; when the detection data is qualified, proceed to the next step;

[0055] Step 5: Detect the transmission hardware fittings in the detected area through the defect detection model. When the detection result is qualified, no operation is performed; when the detection result is unqualified, record the corresponding unqualified area;

[0056] The working method of the defect detection model includes:

[0057] Obtain the collected image of the processed transmission hardware fittings, input the collected image into the defect detection model, analyze the input collected image, and obtain the detection results, including qualified detection and unqualified detection, as well as the unqualified area corresponding to the unqualified detection.

[0058] Step Six: Sort the unqualified transmission hardware fittings;

[0059] The method for sorting the unqualified transmission hardware fittings includes:

[0060] Construct a transmission hardware fitting model, divide different sorting areas in the transmission hardware fitting model, and set corresponding transmission paths and storage locations for each sorting area; obtain the unqualified area corresponding to the transmission hardware fitting with defects, match the corresponding sorting area, and according to the matched sorting area, transmit the transmission hardware fitting with defects to the storage location through the corresponding transmission path.

[0061] The sorting areas are divided by the expert group according to the subsequent processing methods of defects in different positions of the transmission hardware fittings. For example, if there are defects in some positions, but the defects can be overcome by replacing the corresponding accessories, there is no need to completely discard them; if the defects in some positions cause the entire transmission hardware fitting to be unusable, different sorting areas can be defined according to the actual subsequent production needs, which is convenient for subsequent separate processing of the defective transmission hardware fittings, improving production efficiency and resource utilization rate.

[0062] The working principle of the present invention:

[0063] Obtain the production line assembly drawing of the transmission hardware fittings; identify the production line assembly drawing of the transmission hardware fittings, divide the detection area, set a housing device in the detection area, and a multi-color light source is arranged inside the housing device. Identify the type, model and working environment of the transmission hardware fittings to be detected, set a working simulation module inside the housing device according to the working environment of the transmission hardware fittings, set an image acquisition module inside the housing device, obtain the current detection environment, assign the obtained type, model, working environment of the transmission hardware fittings and the current detection environment and convert them into matching vectors, establish a defect detection solution library, input the matching vectors into the defect detection solution library, match the corresponding defect detection solution, and complete the setting of the detection device for the transmission hardware fittings according to the matched defect detection solution.

[0064] Obtain several representative images of transmission small fittings, perform representative image processing to obtain several groups of training sets and test sets; construct a convolutional neural network for defect detection of transmission small fittings, and then train and test the convolutional neural network through the obtained training sets and test sets. Mark the successfully tested convolutional neural network as a defect detection model; obtain the detection data of the working simulation module in the detection device, determine whether the detection data is qualified. When the detection data is unqualified, record the corresponding unqualified area; when the detection data is qualified, use the defect detection model to detect the transmission small fittings in the detected area. When the detection result is qualified, no operation is performed; when the detection result is unqualified, record the corresponding unqualified area; sort the unqualified transmission small fittings.

[0065] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A defect detection method for transmission hardware fittings based on a cascaded network, characterized in that, The specific methods include: Step 1: Obtain the assembly drawings of the production line of transmission line hardware fittings; set up the detection device for transmission line hardware fittings. Among them, the method for setting up the detection device for transmission line hardware fittings includes: Identify the assembly drawings of the production line of transmission line hardware fittings, divide the detection area, set up a housing device in the detection area, and install a multi-color light source inside the housing device. Identify the types, models, and working environments of the transmission line hardware fittings to be detected, set up a working simulation module inside the housing device according to the working environment of the transmission line hardware fittings, set up an image acquisition module inside the housing device, obtain the current detection environment, assign the obtained types, models, working environments of the transmission line hardware fittings, and the current detection environment and convert them into matching vectors, establish a defect detection solution library, input the matching vectors into the defect detection solution library, match the corresponding defect detection solution, and supplement the detection device according to the matched defect detection solution. Step 2: Obtain several representative images of the transmission line hardware fittings, perform representative image processing, and obtain several groups of training sets and test sets. Step 3: Construct a convolutional neural network for defect detection of transmission line hardware fittings, and then train and test the convolutional neural network through the obtained training sets and test sets, and mark the successfully tested convolutional neural network as a defect detection model. Step 4: Obtain the detection data of the working simulation module in the detection device, and determine whether the detection data is qualified. When the detection data is unqualified, record the corresponding unqualified area; when the detection data is qualified, proceed to the next step. Step 5: Detect the transmission line hardware fittings passing through the detection area through the defect detection model. When the detection result is qualified, no operation is performed; when the detection result is unqualified, record the corresponding unqualified area. Step 6: Sort the unqualified transmission line hardware fittings.

2. The method for detecting defects of transmission hardware based on a cascaded network according to claim 1, wherein, The working method of the working simulation module includes: Obtain the working environment of the transmission line hardware fittings, set up the corresponding voltage simulation environment according to the obtained working environment, and establish a corresponding simulation matching table; identify the working environment of the current transmission line hardware fittings, match the corresponding voltage simulation environment from the simulation matching table, detect the current transmission line hardware fittings according to the matched voltage simulation environment, and generate an image acquisition signal during the detection process and send it to the image acquisition module.

3. A method for detecting defects in small transmission hardware based on a cascaded network according to claim 2, characterized in that, The working method of the image acquisition module includes: When receiving the image acquisition signal, control the multi-color light source inside the housing device to turn on the corresponding color light source; perform image acquisition, mark it as the initial acquisition image, and perform background segmentation on the initial acquisition image to obtain the corresponding acquisition image of the transmission line hardware fittings.

4. A method for detecting defects of small transmission hardware based on a cascaded network according to claim 1, characterized in that, The method for establishing the defect detection solution library includes: Obtain the target information of the transmission line hardware fittings, where the target information includes the types, models, and working environments of the transmission line hardware fittings, classify the obtained target information to obtain classified data, formulate a defect detection solution based on the housing device according to the classified data, extract the characteristic vectors of the defect detection solution, establish a first database, store the set defect detection solution in the first database, establish a vector space unit in the first data, map the extracted characteristic vectors into the vector space unit, and mark the current first database as the defect detection solution library.

5. A method for detecting defects of small transmission hardware based on a cascaded network according to claim 1, characterized in that The method for representative image processing includes: Performing super-resolution on the representative image based on the image super-resolution algorithm, marking the super-resolved representative image as the super-resolved image, marking the qualified areas and defective areas in the super-resolved image, and marking the corresponding position information on the defective areas. Segmenting the super-resolved image based on a preset size to obtain a number of segmented images, identifying the marked information on the segmented images, and integrating them into a training set and a test set.

6. The method for detecting defects of transmission hardware based on a cascaded network according to claim 1, wherein The method for sorting out unqualified transmission small metal fittings includes: Constructing a transmission small metal fitting model, dividing different sorting areas in the transmission small metal fitting model, and setting corresponding transmission paths and storage locations for each sorting area; obtaining the unqualified areas corresponding to the transmission small metal fittings with defects, matching the corresponding sorting areas, and transporting the transmission small metal fittings with defects to the storage locations through the corresponding transmission paths according to the matched sorting areas.

7. A defect detection method for transmission hardware fittings based on a cascaded network according to claim 1, characterized in that, The housing device is used to block light and isolate the outside.

Images