Pressing plate state intelligent identification device

Abstract

A pressing plate state intelligent identification device disclosed by the present invention comprises a hand-held terminal main body, an image collection module, an image binding identification module and a local area network communication module, the image collection module is used for collecting an image of a receiving surface on a pressing plate, and the image binding identification module is used for identifying the pressing plate through an algorithm and identifying and homing the same element corresponding to a two-dimensional code. The local area network communication module is used for enabling the handheld terminal main body to be connected with an in-station local area network by utilizing a 5G signal to transmit a signal, and utilizing a server to share identification computing power; according to the method, a mode of handheld terminal main body shooting and two-dimensional code label reading is adopted, and a screen cabinet pressing plate state recognition model is constructed based on a deep learning model by relying on an artificial intelligence algorithm. Carding recording of equipment information and pressing plate states is completed in the inspection process, and a pressing plate state database is formed. And correct on-off states of the pressing plate under different operation contents are learned through a database, and are intelligently pushed to workers to assist the workers in completing related operations, so that the problem of detection mistakes and omissions is prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment, and in particular to a device for intelligently identifying the state of a pressure plate. Background Art

[0002] The relay board is an important component of the relay protection device. The relay protection device is an automatic device. When an electrical component in the power system (such as a generator, transformer, transmission line, etc.) fails or operates abnormally, it can act quickly and selectively remove the faulty component from the power system, or send a signal to ensure that the fault-free part of the power system continues to operate normally. The relay board is mainly used to realize the logical function of relay protection. It determines whether to trigger the protection action by collecting, comparing, calculating and judging the input signal (such as current, voltage and other electrical quantities). In order to promote the safe, stable and efficient operation of the substation, it is necessary to effectively improve the operation and maintenance level of the secondary equipment, especially the inspection level of the relay protection device;

[0003] The existing secondary equipment pressure plate status monitoring is still mainly completed by manual inspection, and due to visual fatigue, memory confusion and other reasons, manual monitoring will lead to the phenomenon of missed operation and wrong operation of the pressure plate. Moreover, when the pressure plate is misoperated due to uncontrollable factors, it can only be discovered and restored after the relevant accident occurs, which seriously endangers the safety and stability of the power system. Therefore, the present invention proposes a pressure plate status intelligent identification device to solve the problems existing in the prior art. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to propose a device for intelligently identifying the status of a pressure plate. The device adopts a mode of shooting with a handheld terminal body and reading a QR code label, relies on an artificial intelligence algorithm, and builds a screen cabinet pressure plate status recognition model based on a deep learning model. During the inspection process, the equipment information and pressure plate status are sorted and recorded to form a pressure plate status database. The correct status of the pressure plate under different operation contents is learned through the database, and intelligent push is given to the staff to assist them in completing related operations, which effectively solves the problem of manual detection of the pressure plate status, which is prone to detection errors.

[0005] To achieve the purpose of the present invention, the present invention is implemented through the following technical solutions: a pressure plate state intelligent recognition device, including a handheld terminal body, an image collection module, an image binding recognition module and a local area network communication module, the image collection module is used to collect images facing the pressure plate, the image binding recognition module is used to identify the pressure plate and the corresponding two-dimensional code through an algorithm to return the same element, the image binding recognition module includes a sample training module, a two-dimensional code label module and an intelligent recognition module, the sample training module is used to input the characteristic image of the target category to build a sample library, the two-dimensional code label module is used to bind with different pressure plate position images, and judge the operation state of the equipment group in combination with the pressure plate insertion and withdrawal state group, and the intelligent recognition module is used to judge the operation state of the equipment group based on the trained YoLo The v5 algorithm identifies the insertion and retraction morphological features and QR code ID of the protective pressure plate, and binds and identifies the pressure plate and the corresponding ID QR code together, and confirms the operating status and provides feedback based on the opening and closing results of the pressure plate and the corresponding ID. The LAN communication module is used to use 5G signals to connect the handheld terminal body to the local area network within the station to transmit signals, and use the server to share the recognition computing power. The LAN communication module includes an identification server module and a storage server module.

[0006] A further improvement is that the method for building a sample library in the sample training module includes the following steps:

[0007] Step 1: Load data. Load the training set data into memory through a custom data loader. The data loader will provide the data to the model for training in batches according to the set batch size.

[0008] Step 2: Network forward propagation, the input image is forward propagated through the YoLo v5 network, and the network outputs multiple feature maps, each of which corresponds to a target category;

[0009] Step 3: Feature map separation: divide the feature map into grids, predict a fixed number of bounding boxes for each grid, and make category predictions for each bounding box;

[0010] Step 4: Non-maximum suppression: Perform non-maximum suppression on the bounding box in step 2 to eliminate redundant prediction results.

[0011] Step 5: Calculate the loss. Calculate the loss function based on the true label and the predicted result, including classification loss and bounding box coordinate loss.

[0012] Step 6: Back propagation, based on the loss calculated in step 5, use the loss function to calculate the gradient and update the network parameters through back propagation;

[0013] Step 7: Parameter optimization, use the optimization algorithm to optimize the network parameters to minimize the loss function;

[0014] Step 8: Output the results. After multiple rounds of iterations, the optimized prediction results are finally output.

[0015] Further improvements are: the image collection module includes a shooting module, a dark area fill light module and an image cutout module, the shooting module is used to take photos of the pressure plate's insertion and retraction status and the QR code label of the corresponding ID, the dark area fill light module is used to provide auxiliary lighting for quickly taking photos when the external environment light is relatively dim, the dark area fill light module includes a brightness sensing unit and a fill light unit, and the image cutout module is used to cut out the pressure plate area in the picture from the complex background to prevent the background image of the photo from affecting recognition.

[0016] A further improvement is that the brightness sensing unit includes a light-sensitive sensor for sending a signal to start fill light after the external environment brightness drops below a set value.

[0017] A further improvement is that the fill light unit includes a fill light for irradiating light to the front of the device to increase the brightness and facilitate shooting and reading labels.

[0018] A further improvement is that the recognition server module is used to share the recognition computing power of the handheld terminal body for the captured pictures, thereby reducing the workload of the handheld terminal body.

[0019] A further improvement is that the storage server module is used to receive and store the pressure plate pictures and QR code ID data taken by the handheld terminal body, thereby alleviating the storage pressure of the handheld terminal body.

[0020] The beneficial effects of the present invention are as follows: the present invention adopts the mode of shooting with a handheld terminal and reading a QR code label, relying on an artificial intelligence algorithm, and constructing a screen cabinet pressure plate state recognition model based on a deep learning model. During the inspection process, the equipment information and pressure plate state are sorted and recorded to form a pressure plate state database. The correct state of the pressure plate under different operation contents is learned through the database, and intelligent push is sent to the staff to assist them in completing related operations, which effectively solves the problem of manual detection of the pressure plate state, which is prone to detection errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a system architecture diagram of the present invention;

[0022] Figure 2 This is a diagram of the dark area fill light module architecture of the present invention. DETAILED DESCRIPTION

[0023] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0024] according to Figure 1As shown, this embodiment provides a device for intelligently identifying the state of a pressure plate, including a handheld terminal body, an image collection module, an image binding recognition module and a local area network communication module. The image collection module is used to collect images facing the pressure plate, and the image binding recognition module is used to identify the pressure plate and the corresponding two-dimensional code through an algorithm to return the pressure plate recognition and the corresponding two-dimensional code recognition to the same element. The image binding recognition module includes a sample training module, a two-dimensional code label module and an intelligent recognition module. The sample training module is used to input feature images of the target category to build a sample library, and the two-dimensional code label module is used to bind with different pressure plate position images, and judge the operation state of the equipment group in combination with the pressure plate insertion and withdrawal state group. The intelligent recognition module is used to calculate the operation state of the equipment group based on the trained YoLov5 algorithm. The method can identify the morphological features and QR code ID of the protective pressure plate. The investment and withdrawal forms include two states: investment and withdrawal. The upright form is the investment state pressure plate, and the inclined form is the withdrawal state pressure plate. When identifying, the direction angles of the investment and withdrawal states of the pressure plate are obviously different. The direction angle is used as the discrimination feature, and the pressure plate and the corresponding ID QR code are bound and identified together. The operating state is confirmed and feedback is given according to the opening and closing results of the pressure plate and the corresponding ID. The algorithm is used to identify and confirm the operating state of the pressure plate equipment, correctly and effectively identify the investment and withdrawal state of the pressure plate, assist in completing the inspection, effectively improve the inspection efficiency and accuracy, and solve the problem that traditional manual inspection and monitoring are easily affected by human errors, resulting in missed operations and misoperations of the pressure plate.

[0025] The method for building a sample library in the sample training module includes the following steps:

[0026] Step 1: Load data and load the training set data into memory through a custom data loader. The data loader will provide the data to the model for training in batches according to the set batch size;

[0027] Step 2: Network forward propagation, the input image is forward propagated through the YoLo v5 network, and the network outputs multiple feature maps, each of which corresponds to a target category;

[0028] Step 3: Feature map separation: divide the feature map into grids, predict a fixed number of bounding boxes for each grid, and make category predictions for each bounding box;

[0029] Step 4: Non-maximum suppression: Perform non-maximum suppression on the bounding box in step 2 to eliminate redundant prediction results.

[0030] Step 5: Calculate the loss. Calculate the loss function based on the true label and the predicted result, including classification loss and bounding box coordinate loss.

[0031] Step 6: Back propagation, based on the loss calculated in step 5, use the loss function to calculate the gradient and update the network parameters through back propagation;

[0032] Step 7: Parameter optimization, use the optimization algorithm to optimize the network parameters to minimize the loss function;

[0033] Step 8: Output the results. After multiple rounds of iterations, the optimized prediction results are finally output.

[0034] The image collection module includes a shooting module, a dark area fill light module and an image cutout module. The shooting module is used to shoot photos of the pressure plate's insertion and withdrawal status and the QR code label of the corresponding ID. The dark area fill light module is used to provide auxiliary lighting for quickly shooting photos when the external environment is dim. The dark area fill light module includes a brightness sensing unit and a fill light unit. The image cutout module is used to cut out the pressure plate area in the picture from the complex background to prevent the background image of the photo from affecting recognition. The brightness sensing unit includes a photosensitive sensor for sending a signal to start the fill light after the external environment brightness drops below a set value. The fill light unit includes a fill light for irradiating light to the front of the device to increase the brightness for easy shooting and reading of labels. When collecting images, the handheld terminal body is used to aim at the pressure plate in different insertion and withdrawal states with the shooting module to take photos and collect photos. The photosensitive sensor and the fill light cooperate to provide lighting when the external environment brightness is low to ensure the clarity and completeness of the image, while avoiding the background pattern from affecting the algorithm analysis and judgment.

[0035] The LAN communication module is used to use 5G signals to connect the handheld terminal body to the local area network within the station to transmit signals, and use the server to share the recognition computing power. The LAN communication module includes an identification server module and a storage server module. The identification server module is used to share the recognition computing power of the handheld terminal body for the captured pictures, thereby reducing the workload of the handheld terminal body. The storage server module is used to receive and store the pressure plate pictures and QR code ID data taken by the handheld terminal body, thereby reducing the storage pressure of the handheld terminal body.

[0036] Before using the intelligent recognition device for the state of the pressure plate, the target images with different discriminant features are first input into the sample library and the YoLo v5 algorithm is used for training and iterative recognition. When the staff conducts patrol inspection on the pressure plate, the handheld terminal body is aimed at the pressure plate, and the shooting module is used to shoot. The dark part fill light module ensures the brightness of the photographed photo. When the light sensor senses that the brightness of the external environment is dark, the fill light is started to illuminate to ensure the clarity of the picture. The pressure plate is cut out from the redundant complex background by the image cutout module. The trained algorithm is used to detect the position of the pressure plate and identify the QR code corresponding to the nearest position of the pressure plate to obtain the corresponding ID. The operation state is confirmed and feedback is given according to the opening and closing result of the pressure plate and the corresponding ID. When identifying, the corresponding QR code recognition and pressure plate recognition are classified as the same element. The pressure plate frame is responsible for identifying the insertion and withdrawal state. The QR code recognition is used to bind the corresponding ID. The storage server module is used as a database to store the corresponding ID. The operation state of the equipment group is judged according to the QR code ID and the pressure plate insertion and withdrawal state group.

[0037] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A device for intelligently identifying the state of a pressing plate, characterized in that: It includes a handheld terminal body, an image collection module, an image binding recognition module and a local area network communication module. The image collection module is used to collect images facing the pressure plate. The image binding recognition module is used to identify the pressure plate and the corresponding two-dimensional code through an algorithm to return them to the same element. The image binding recognition module includes a sample training module, a two-dimensional code label module and an intelligent recognition module. The sample training module is used to input feature images of the target category to build a sample library. The two-dimensional code label module is used to bind with different pressure plate position images, and judge the operation state of the equipment group in combination with the pressure plate insertion and withdrawal state group. The intelligent recognition module is used to identify the insertion and withdrawal morphological features and two-dimensional code ID of the protective pressure plate based on the trained YoLo v5 algorithm, and at the same time, the pressure plate and the corresponding ID two-dimensional code are jointly bound and recognized, and the operation state is confirmed and feedback is given according to the opening and closing result of the pressure plate and the corresponding ID. The local area network communication module is used to use 5G signals to connect the handheld terminal body to the local area network in the station to transmit signals, and use the server to share the recognition computing power. The local area network communication module includes an identification server module and a storage server module.

2. The intelligent identification device for a pressing plate state according to claim 1, characterized in that: The method for building a sample library in the sample training module comprises the following steps: Step 1: Load data. Load the training set data into memory through a custom data loader. The data loader will provide the data to the model for training in batches according to the set batch size. Step 2: Network forward propagation, the input image is forward propagated through the YoLo v5 network, and the network outputs multiple feature maps, each of which corresponds to a target category; Step 3: Feature map separation: divide the feature map into grids, predict a fixed number of bounding boxes for each grid, and make category predictions for each bounding box; Step 4: Non-maximum suppression: Perform non-maximum suppression on the bounding box in step 2 to eliminate redundant prediction results. Step 5: Calculate the loss. Calculate the loss function based on the true label and the predicted result, including classification loss and bounding box coordinate loss. Step 6: Back propagation, based on the loss calculated in step 5, use the loss function to calculate the gradient and update the network parameters through back propagation; Step 7: Parameter optimization, use the optimization algorithm to optimize the network parameters to minimize the loss function; Step 8: Output the results. After multiple rounds of iterations, the optimized prediction results are finally output.

3. The intelligent identification device for a pressing plate state according to claim 1, characterized in that: The image collection module includes a shooting module, a dark area fill light module and an image cutout module. The shooting module is used to take photos of the QR code label of the pressure plate's insertion and retraction status and the corresponding ID. The dark area fill light module is used to provide auxiliary lighting for quickly taking photos when the external ambient light is relatively dim. The dark area fill light module includes a brightness sensing unit and a fill light unit. The image cutout module is used to cut out the pressure plate area in the picture from the complex background to prevent the background image of the photo from affecting recognition.

4. The intelligent identification device for a pressing plate state according to claim 3, characterized in that: The brightness sensing unit includes a light-sensitive sensor, which is used to send a signal to start fill light after the external environment brightness drops below a set value.

5. The intelligent identification device for a pressing plate state according to claim 3, characterized in that: The fill light unit includes a fill light for irradiating light to the front of the device to increase the brightness and facilitate shooting and label reading.

6. The intelligent identification device for a pressing plate state according to claim 1, characterized in that: The recognition server module is used to share the recognition computing power of the handheld terminal body for the captured pictures, thereby reducing the workload of the handheld terminal body.

7. The intelligent identification device for a pressing plate state according to claim 1, characterized in that: The storage server module is used to receive and store the pressing plate pictures and QR code ID data taken by the handheld terminal body, thereby alleviating the storage pressure of the handheld terminal body.

Images