Strip steel exiting detection method, device, equipment and medium

Abstract

The invention discloses a strip steel exit detection method, device and equipment and a medium, and the method comprises the steps: responding to a strip steel detection starting signal, and controlling image equipment to collect image data of a target area according to a preset frequency; the collected image data are input into a pre-trained target residual network model, a judgment result output by the target residual network model is obtained, and the judgment result represents whether strip steel exists at a target position in the target area or not; and when the judgment result is that the strip steel does not exist at the target position, generating a reminding signal that the strip steel is quitted. It can be seen that whether the strip steel at the strip steel inlet and outlet position of the unbundling machine is completely extracted or not is determined in an image data analysis mode, the mode that whether the strip steel completely exits from a five-roller straightening machine and an unbundling device or not is adopted in the related technology is avoided, and the accuracy of the strip steel detection result is improved.

Description

Technical Field

[0001] The invention relates to the field of cold rolling technology, and in particular to a strip steel exit detection method, device, equipment and medium. Background Art

[0002] There is an operation involved in the cold rolling process, that is, after the debundling machine debundles the strip steel coil, the strip head is sent to the transverse shear to shear the strip head. After the strip head shearing is completed, the strip steel returns to the five-roll straightening machine and the debundling device. In this process, it is necessary to determine whether the strip steel is completely withdrawn. Only after it is determined that the strip steel is completely withdrawn can the walking beam be controlled to send the steel coil away.

[0003] In the related art, photoelectric tubes are mainly used to detect whether the strip steel has completely exited the five-roll straightener and the debundling device. However, the photoelectric tubes are easily polluted by broken straps, iron oxide powder or dust, resulting in inaccurate detection results. Therefore, how to improve the detection accuracy of whether the strip steel has completely exited the five-roll straightener and the debundling device is a problem that needs to be solved urgently. Summary of the invention

[0004] The embodiments of the present application provide a strip exit detection method, device, equipment and medium, which solves the technical problem that in the prior art, photoelectric tubes are mainly used to detect whether the strip has completely exited from the five-roll straightening machine and the debundling device, but the photoelectric tubes are easily contaminated and interfered by broken straps, iron oxide powder or dust, resulting in inaccurate detection results. The technical effect of improving the detection accuracy of whether the strip has completely exited from the five-roll straightening machine and the debundling device is achieved.

[0005] In a first aspect, the present application provides a strip exit detection method, the method comprising:

[0006] In response to a strip steel detection start signal, controlling an imaging device to collect image data of a target area at a preset frequency;

[0007] The collected image data is input into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel in a target position in the target area;

[0008] When the judgment result is that there is no strip steel at the target position, a reminder signal is generated indicating that the strip steel has exited.

[0009] Furthermore, after obtaining the judgment result of the target residual network model output, the method further includes:

[0010] When the judgment result is that there is a steel strip at the target position, the process returns to the step of controlling the imaging device to collect image data of the target area at a preset frequency until the judgment result is that there is no steel strip at the target position.

[0011] Furthermore, before inputting the collected image data into the pre-trained target residual network model, the method further includes:

[0012] Construct the initial residual network model based on the convolutional layer, maximum pooling layer, attention mechanism layer, average pooling layer and fully connected layer;

[0013] The initial residual network model is trained to obtain the target residual network model.

[0014] Furthermore, based on the convolutional layer, the maximum pooling layer, the attention mechanism layer, the average pooling layer and the fully connected layer, an initial residual network model is constructed, including:

[0015] According to the convolution layer and the maximum pooling layer, the first group of blocks is constructed;

[0016] According to the convolution layer and the attention mechanism layer, the second group of blocks is constructed;

[0017] Construct the third group of blocks based on the attention mechanism layer, average pooling layer, and fully connected layer;

[0018] An initial residual network model is constructed according to the first component block, the third component block and at least four groups of second component blocks, each group of second component blocks including at least one second component block.

[0019] Further, when there are 4 groups of second component blocks, the number of second component blocks included in each group of second component blocks is 3, 4, 6, and 3 respectively.

[0020] Further, an initial residual network model is constructed according to the first component block, the third component block and at least four groups of second component blocks, including:

[0021] The first component block, at least four groups of the second component blocks, and the third component block are connected in sequence to construct an initial residual network model.

[0022] Furthermore, the attention mechanism layer is an ECA attention mechanism layer.

[0023] In a second aspect, the present application provides a strip exit detection device, the device comprising:

[0024] An image acquisition module, used to control the image device to acquire image data of the target area at a preset frequency in response to a strip steel detection start signal;

[0025] A judgment module is used to input the collected image data into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel in the target position in the target area;

[0026] The exit module is used to generate a reminder signal that the strip has been exited when the judgment result is that there is no strip at the target position.

[0027] Furthermore, the loop module is used to:

[0028] After obtaining the judgment result output by the target residual network model, when the judgment result is that there is a strip steel at the target position, return to the step of controlling the image device to collect image data of the target area at a preset frequency until the judgment result is that there is no strip steel at the target position.

[0029] Furthermore, the model building module is used to:

[0030] Before inputting the collected image data into the pre-trained target residual network model, an initial residual network model is constructed according to the convolution layer, the maximum pooling layer, the attention mechanism layer, the average pooling layer, and the fully connected layer;

[0031] The initial residual network model is trained to obtain the target residual network model.

[0032] Furthermore, the model building module is used to:

[0033] According to the convolution layer and the maximum pooling layer, the first group of blocks is constructed;

[0034] According to the convolution layer and the attention mechanism layer, the second group of blocks is constructed;

[0035] Construct the third group of blocks based on the attention mechanism layer, average pooling layer, and fully connected layer;

[0036] An initial residual network model is constructed according to the first component block, the third component block and at least four groups of second component blocks, each group of second component blocks including at least one second component block.

[0037] Further, when there are 4 groups of second component blocks, the number of second component blocks included in each group of second component blocks is 3, 4, 6, and 3 respectively.

[0038] Furthermore, the model building module is used to:

[0039] The first component block, at least four groups of the second component blocks, and the third component block are connected in sequence to construct an initial residual network model.

[0040] Furthermore, the attention mechanism layer is an ECA attention mechanism layer.

[0041] In a third aspect, the present application provides an electronic device, including:

[0042] processor;

[0043] a memory for storing processor-executable instructions;

[0044] Wherein, the processor is configured to execute to implement a strip exit detection method as provided in the first aspect.

[0045] In a fourth aspect, the present application provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to implement a strip exit detection method as provided in the first aspect.

[0046] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0047] In the embodiment of the present application, in response to the strip steel detection start signal, the image device is controlled to collect image data of the target area at a preset frequency; the collected image data is input into the pre-trained target residual network model to obtain the judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel at the target position in the target area; when the judgment result is that there is no strip steel at the target position, a reminder signal that the strip steel has been withdrawn is generated. It can be seen that the embodiment of the present application determines whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn by image data analysis, avoiding the method of detecting whether the strip steel is completely withdrawn from the five-roller straightening machine and the debundling device in the related art, and also avoiding the problem that the photoelectric tube may output the result that the strip steel has been completely withdrawn due to the interference of broken straps, iron oxide powder or dust pollution, resulting in inaccurate detection results, and thus the embodiment of the present application improves the accuracy of the strip steel detection result. After improving the accuracy of the detection result of whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn, the probability of the walking beam rolling coil pulling the steel coil crooked and the probability of causing the coil to turn over are reduced, and the production stability is improved. At the same time, it also reduces the number of times the crane is used, thereby reducing the possibility of crooked and slanted lifting scenes and reducing safety hazards. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 A schematic diagram of a process flow of a strip steel exit detection method provided in an embodiment of the present application;

[0050] Figure 2 A schematic diagram of a target area and a target position provided in an embodiment of the present application;

[0051] Figure 3A schematic diagram of the residual unit structure provided in an embodiment of the present application;

[0052] Figure 4 A schematic diagram of the structure of the target residual network model provided in the embodiment of the present application;

[0053] Figure 5 A schematic diagram of the network structure of the channel attention mechanism provided in an embodiment of the present application;

[0054] Figure 6 A schematic diagram of the structure of the input and output of the target residual network model provided in the embodiment of the present application;

[0055] Figure 7 A diagram of two judgment results provided in an embodiment of the present application;

[0056] Figure 8 A schematic diagram of the structure of a strip steel exit detection device provided in an embodiment of the present application;

[0057] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] The embodiment of the present application provides a strip exit detection method, which solves the technical problem that in the prior art, photoelectric tubes are mainly used to detect whether the strip has completely exited the five-roll straightening machine and the debundling device. However, the photoelectric tubes are easily interfered by broken straps, iron oxide powder or dust pollution, resulting in inaccurate detection results.

[0059] The technical solution of the embodiment of the present application is to solve the above technical problems, and the overall idea is as follows:

[0060] In the embodiment of the present application, in response to the strip steel detection start signal, the image device is controlled to collect image data of the target area at a preset frequency; the collected image data is input into the pre-trained target residual network model to obtain the judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel at the target position in the target area; when the judgment result is that there is no strip steel at the target position, a reminder signal that the strip steel has been withdrawn is generated. It can be seen that the embodiment of the present application determines whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn by image data analysis, avoiding the method of detecting whether the strip steel is completely withdrawn from the five-roller straightening machine and the debundling device in the related art, and also avoiding the problem that the photoelectric tube may output the result that the strip steel has been completely withdrawn due to the interference of broken straps, iron oxide powder or dust pollution, resulting in inaccurate detection results, and thus the embodiment of the present application improves the accuracy of the strip steel detection result. After improving the accuracy of the detection result of whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn, the probability of the walking beam rolling coil pulling the steel coil crooked and the probability of causing the coil to turn over are reduced, and the production stability is improved. At the same time, it also reduces the number of times the crane is used, thereby reducing the possibility of crooked and slanted lifting scenes and reducing safety hazards.

[0061] 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.

[0062] First of all, the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0063] There is an operation involved in the cold rolling process, that is, after the debundling machine debundles the strip steel coil, the strip head is sent to the transverse shear to shear the strip head. After the strip head shearing is completed, the strip steel returns to the five-roll straightening machine and the debundling device, and then the walking beam is controlled to send the steel coil away. In this process, it is necessary to determine whether the strip steel is completely withdrawn. Only after it is determined that the strip steel is completely withdrawn can the walking beam be controlled to send the steel coil away.

[0064] In the related art, photoelectric tubes are mainly used to detect whether the strip is completely withdrawn from the five-roll straightening machine and the debundling device. However, photoelectric tubes are easily interfered by broken straps, iron oxide powder or dust pollution. When the strip is not completely withdrawn, the photoelectric tubes may output the result that the strip has been completely withdrawn due to interference from broken straps, iron oxide powder or dust pollution, resulting in inaccurate detection results.

[0065] When the strip is not completely withdrawn, the photoelectric tube detects that the strip has been completely withdrawn, and the walking beam will pull the steel coil crooked (that is, deviate from the horizontal line), and in serious cases, it will cause rollover (rollover means that the entire steel coil is completely overturned). In addition, the crane cannot be driven directly above the debundling position. In the process of handling the crooked steel coil, it will form a crooked and oblique lifting, which poses a great safety hazard and affects the operation of the unit. It can be seen that the accuracy of the detection result of whether the strip is completely withdrawn has a greater impact on subsequent processing. Therefore, how to improve the detection accuracy of whether the strip is completely withdrawn from the five-roll straightening machine and the debundling device is a problem that needs to be solved urgently.

[0066] In order to solve the above problems, the present application provides the following embodiments: Figure 1 A strip steel exit detection method is shown, and the method includes steps S11 to S13.

[0067] Step S11, in response to a strip steel detection start signal, controlling an imaging device to collect image data of a target area at a preset frequency;

[0068] Step S12, inputting the collected image data into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, wherein the judgment result indicates whether there is a strip steel at a target position in the target area;

[0069] Step S13, when the judgment result is that the strip steel does not exist at the target position, a reminder signal that the strip steel has exited is generated.

[0070] A strip exit detection method provided in an embodiment of the present application can be executed by a host computer of a cold rolling production line, or by a controller of a walking beam, and the embodiment of the present application does not impose any limitation on this.

[0071] Step S11, in response to the strip steel detection start signal, controlling the imaging device to collect image data of the target area at a preset frequency.

[0072] The strip steel detection start signal can be generated by starting the debundling machine to generate the strip steel detection start signal. After receiving the strip steel detection start signal, the imaging device can be controlled to collect image data of the target area at a preset frequency.

[0073] The preset frequency can be set according to actual needs. For example, when the strip is drawn out at a faster speed, a higher preset frequency is set, and when the strip is drawn out at a slower speed, a lower preset frequency is set.

[0074] The target area is the area including the gap between the strip steel coil and the debaler. Figure 2 As shown in the figure, the red box is the gap area between the strip steel coil and the debundling machine, which is recorded as the target position in the embodiment of the present application. Figure 2 It refers to the target area.

[0075] Each time a piece of image data of the target area is collected, conventional image preprocessing may be performed thereon, and then step S12 is continued.

[0076] Step S12, inputting the collected image data into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, wherein the judgment result represents whether there is a strip steel in the target position in the target area.

[0077] The collected image data is input into a pre-trained target residual network model, which can identify whether there is still strip steel at the target position in the image data. When there is strip steel, step S14 is executed, and when there is no strip steel, step S13 is executed.

[0078] Among them, the residual network model is also called the ResNet network model. ResNet includes different types, the most important of which are ResNet-18 and ResNet-50. These two types are the most advanced and common in image classification. ResNet50 has a deeper network structure than ResNet18 and has more advantages in feature extraction of image classification. Therefore, the embodiment of this application uses the ResNet-50 model as the basic model for improvement. Figure 3 As shown in FIG, it is a schematic diagram of the residual unit structure, where weaklayer is the weight layer, F(x) is the weight function expression, the arrow line on the right is the shortcut connection, and x is the input parameter.

[0079] The ResNet-50 model consists of 48 convolutional layers, 1 MaxPool layer (also known as the maximum pooling layer), an intermediate pooling layer, and a ReLU activation method.

[0080] One feature of ResNet is that it can overcome the overfitting problem. In a feedforward neural network, a hidden layer is included. The hidden layer contains multiple specific neurons that are sufficient to fill the data in a linear manner, and this hidden layer may contain a large number of parameters, so network training becomes associated with a specific amount of data. The overfitting problem causes the accuracy of ResNet in image classification to decrease. The embodiment of the present application obtains a pre-training model based on the ResNet network model, which can overcome the overfitting problem.

[0081] ResNet also successfully overcomes the vanishing gradient problem, where iterative multiplication of derivative values ​​of previous layers results in a decrease in derivative values, which leads to a decrease in network accuracy for deep learning. The idea of ​​ResNet is to reduce the accumulation of layers by adding something called "shortcut connections," which bypass one or more layers in the network structure, thereby reducing errors during data training.

[0082] Before inputting the collected image data into a pre-trained target residual network model, the embodiment of the present application first constructs an initial residual network model based on a convolutional layer, a maximum pooling layer, an attention mechanism layer, an average pooling layer, and a fully connected layer; then the initial residual network model is trained to obtain a target residual network model.

[0083] Specifically, an initial residual network model is constructed according to the convolutional layer, the maximum pooling layer, the attention mechanism layer, the average pooling layer and the fully connected layer, including steps S121 to S124.

[0084] Step S121, constructing a first grouping block according to the convolution layer and the maximum pooling layer;

[0085] Step S122, constructing a second component block according to the convolution layer and the attention mechanism layer;

[0086] Step S123, constructing a third group of blocks according to the attention mechanism layer, the average pooling layer and the fully connected layer;

[0087] Step S124: construct an initial residual network model according to the first component block, the third component block and at least four groups of second component blocks, each group of second component blocks including at least one second component block.

[0088] The first component block, at least four groups of the second component blocks, and the third component block are connected in sequence to construct an initial residual network model.

[0089] The structure of the initial residual network model can be referenced Figure 4 , Conv represents the convolution layer in the neural network, "Conv, 7×7, 64" represents the convolution layer with a convolution kernel size of 7×7 and 64 channels. Max pool represents the maximum pooling layer, ECA block represents the attention mechanism layer, Average pool represents the average pooling layer, and Softmax is the fully connected layer. The entire model contains 4 groups of Building blocks, that is, 4 groups of second component blocks. When there are 4 groups of second component blocks, the number of second component blocks contained in each group of second component blocks can be 3, 4, 6, and 3 respectively. This number selection is a comprehensive consideration when designing the model in order to find a balance between network depth, feature extraction capability, and computational complexity. This number selection can extract low-level features in the early stage, deeply explore complex features in the middle stage, and reduce the computational burden in the final stage, thereby ensuring efficient feature extraction capabilities and model performance to ensure high accuracy of image recognition.

[0090] The attention mechanism layer is the ECA attention mechanism layer. The attention mechanism draws on the selection mechanism of human vision, focusing on information with obvious features and selectively ignoring some minor information to achieve optimal recognition performance. The channel attention mechanism (ECA attention mechanism) is added to the ResNet50 model to improve the model's feature extraction ability for the strip steel image at the debaler. The network structure of the channel attention mechanism is as follows: Figure 5 shown.

[0091] In addition to the ECA attention mechanism, there is also the SE attention mechanism. Regarding the SE attention mechanism, the input feature map is firstly channel compressed, and such compression and dimensionality reduction have an adverse effect on the dependency between learning channels, so the ECA attention mechanism is selected in the embodiment of the present application. The ECA attention mechanism avoids dimensionality reduction, and uses one-dimensional convolution to efficiently implement a global average pooling operation on the input feature map, performs a 1-dimensional convolution operation with a convolution kernel size of k, and obtains the weight w of each channel through the Sigma id activation function, and multiplies the weight with the corresponding element of the original input feature map to obtain the final output feature map.

[0092] From the above, we can see that the idea and operation of the ECA attention mechanism are very simple, and have minimal impact on network processing speed.

[0093] In the model training process, the loss value is obtained by the forward propagation process, and then the back propagation process is entered to obtain the gradient of the parameters. The role of the optimizer is to use the gradient to update the parameters so that the loss is continuously reduced. The Adam (Adaptive Moment Estimation algorithm) optimizer combines the advantages of the gradient descent algorithm of the adaptive learning rate and the momentum gradient descent algorithm, which can adapt to sparse gradients and alleviate the problem of gradient oscillation. When the Adam optimizer updates the loss gradient, it can be based on different parameter adaptive learning rates. In order to better promote model convergence, the ResNet50 model of the present application embodiment introduces cosine annealing (Cos i ne Annea ling) attenuation learning rate in the Adam optimizer to improve the performance of the Ad-am optimizer.

[0094] That is to say, based on the ResNet50 model, the embodiment of the present application integrates the ECA attention mechanism and the Adam optimizer, and introduces an ECA channel attention mechanism layer at the end of each group of residual modules to improve the model's feature extraction ability for the volume head image. The ECA channel attention mechanism is introduced at the end of each building block of ResNet, which can effectively improve the feature extraction ability and enhance the model's discrimination while maintaining a low computational overhead, thereby improving the model's processing effect on complex tasks. In addition, in order to enable the model to converge quickly, the embodiment of the present application uses ImageNet as a pre-training model.

[0095] like Figure 6 As shown, it is a schematic diagram of the structure of the trained target residual network model. The image is input into the first component block, and after being processed by the first component block, the second component block and the third component block, a judgment result of whether the strip steel is completely withdrawn is obtained.

[0096] When the judgment result is that there is no strip steel at the target position, refer to Figure 7 In the right figure, execute step S13. When the judgment result is that there is strip steel at the target position, refer to Figure 7 Left figure, execute step S14.

[0097] Step S13, when the judgment result is that the strip steel does not exist at the target position, a reminder signal that the strip steel has exited is generated.

[0098] When the result of the judgment is that there is no strip steel at the target position, a reminder signal is generated indicating that the strip steel has exited, indicating that the walking beam can be controlled to transport the strip steel. In addition, the imaging device is controlled to sleep or shut down until the next strip steel detection start signal is received, and then steps S11 to S14 are continued.

[0099] Step S14, when the judgment result is that there is a steel strip at the target position, return to the step of controlling the imaging device to collect image data of the target area at a preset frequency until the judgment result is that there is no steel strip at the target position.

[0100] When the judgment result is that there is a steel strip at the target position, it means that the steel strip has not been completely withdrawn, and then the process continues to return to step S11-step S12 until the judgment result is that there is no steel strip at the target position.

[0101] In summary, the embodiment of the present application controls the image device to collect image data of the target area at a preset frequency in response to the strip steel detection start signal; inputs the collected image data into the pre-trained target residual network model to obtain the judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel at the target position in the target area; when the judgment result is that there is no strip steel at the target position, a reminder signal that the strip steel has been withdrawn is generated. It can be seen that the embodiment of the present application determines whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn by image data analysis, avoiding the method of detecting whether the strip steel is completely withdrawn from the five-roller straightening machine and the debundling device in the related art, and also avoiding the problem that the photoelectric tube may output the result that the strip steel has been completely withdrawn due to the interference of broken straps, iron oxide powder or dust pollution, resulting in inaccurate detection results, and thus the embodiment of the present application improves the accuracy of the strip steel detection results. After improving the accuracy of the detection result of whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn, the probability of the walking beam rolling coil pulling the steel coil crooked and the probability of causing the coil to roll over are reduced, and the production stability is improved. At the same time, it also reduces the number of times the crane is used, thereby reducing the possibility of crooked and slanted lifting scenes and reducing safety hazards.

[0102] from Figure 7 It can be seen that the embodiment of the present application uses a deep learning target detection algorithm to identify and judge the image of the strip entrance and exit position of the debaling machine of a certain factory's pickling mill. From the figure, it can be seen that the algorithm is relatively accurate in judging whether the strip is completely extracted from the debaling machine, and the normalized correlation coefficient of the algorithm for strip recognition and detection at the debaling machine is 0.83, which has good robustness.

[0103] A strip steel exit detection method provided in an embodiment of the present application needs to rely on an image device for image acquisition, and then use image preprocessing technology and deep learning technology to infer the image acquired by the camera module, identify the roll head and output the detection result. It is also necessary to rely on a communication module to realize communication between the image device and the controller. The data collected by the image device is displayed in the terminal interface in the form of video, realizing real-time monitoring of the detection screen and providing a visual screen for manual verification of the results. Record the system operation log and provide a text record that can be reviewed for abnormal situations. Implement the image saving function to provide samples for improving the deep learning model, and iterate the model to improve the recognition accuracy. Respond to abnormal power outages and abnormal exits of the server, and automatically restart the system in time. Add a manual unlock button to the terminal interface to unlock the chain when the output result is abnormal, and release it in time to avoid affecting the production rhythm.

[0104] Based on the same inventive concept, the present application embodiment provides the following Figure 8 A strip steel exit detection device is shown, the device comprising:

[0105] An image acquisition module 81 is used to control the image device to acquire image data of the target area at a preset frequency in response to a strip steel detection start signal;

[0106] A judgment module 82 is used to input the collected image data into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel at a target position in the target area;

[0107] The exit module 83 is used to generate a reminder signal that the strip has exited when the judgment result is that the strip does not exist at the target position.

[0108] Furthermore, the loop module is used to:

[0109] After obtaining the judgment result output by the target residual network model, when the judgment result is that there is a strip steel at the target position, return to the step of controlling the image device to collect image data of the target area at a preset frequency until the judgment result is that there is no strip steel at the target position.

[0110] Furthermore, the model building module is used to:

[0111] Before inputting the collected image data into the pre-trained target residual network model, an initial residual network model is constructed according to the convolution layer, the maximum pooling layer, the attention mechanism layer, the average pooling layer, and the fully connected layer;

[0112] The initial residual network model is trained to obtain the target residual network model.

[0113] Furthermore, the model building module is used to:

[0114] According to the convolution layer and the maximum pooling layer, the first group of blocks is constructed;

[0115] According to the convolution layer and the attention mechanism layer, the second group of blocks is constructed;

[0116] Construct the third group of blocks based on the attention mechanism layer, average pooling layer, and fully connected layer;

[0117] An initial residual network model is constructed according to the first component block, the third component block and at least four groups of second component blocks, each group of second component blocks including at least one second component block.

[0118] Further, when there are 4 groups of second component blocks, the number of second component blocks included in each group of second component blocks is 3, 4, 6, and 3 respectively.

[0119] Furthermore, the model building module is used to:

[0120] The first component block, at least four groups of the second component blocks, and the third component block are connected in sequence to construct an initial residual network model.

[0121] Furthermore, the attention mechanism layer is an ECA attention mechanism layer.

[0122] Based on the same inventive concept, the present application embodiment provides the following Fig. 9 An electronic device as shown includes:

[0123] Processor 91;

[0124] A memory 92 for storing instructions executable by the processor 91;

[0125] The processor 91 is configured to execute to implement a strip exit detection method as provided above.

[0126] Based on the same inventive concept, an embodiment of the present application provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by the processor 91 of the electronic device, the electronic device can execute a strip exit detection method as provided above.

[0127] Since the electronic device introduced in this embodiment is an electronic device used to implement the information processing method in the embodiment of the present application, based on the information processing method introduced in the embodiment of the present application, a person skilled in the art can understand the specific implementation of the electronic device of the present embodiment and its various variations, so how the electronic device implements the method in the embodiment of the present application is not described in detail here. As long as a person skilled in the art implements the electronic device used by the information processing method in the embodiment of the present application, it belongs to the scope of protection of this application.

[0128] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0129] In the embodiment of the present application, in response to the strip steel detection start signal, the image device is controlled to collect image data of the target area at a preset frequency; the collected image data is input into the pre-trained target residual network model to obtain the judgment result output by the target residual network model, and the judgment result represents whether there is a strip steel at the target position in the target area; when the judgment result is that there is no strip steel at the target position, a reminder signal that the strip steel has been withdrawn is generated. It can be seen that the embodiment of the present application determines whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn by image data analysis, avoiding the method of detecting whether the strip steel is completely withdrawn from the five-roller straightening machine and the debundling device in the related art, and also avoiding the problem that the photoelectric tube may output the result that the strip steel has been completely withdrawn due to the interference of broken straps, iron oxide powder or dust pollution, resulting in inaccurate detection results, and thus the embodiment of the present application improves the accuracy of the strip steel detection result. After improving the accuracy of the detection result of whether the strip steel at the strip steel inlet and outlet position of the debundling machine is completely withdrawn, the probability of the walking beam rolling coil pulling the steel coil crooked and the probability of causing the coil to turn over are reduced, and the production stability is improved. At the same time, it also reduces the number of times the crane is used, thereby reducing the possibility of crooked and slanted lifting scenes and reducing safety hazards.

[0130] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0131] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0132] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0133] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0134] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0135] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A strip exit detection method, characterized in that: The method comprises: In response to a strip steel detection start signal, controlling an imaging device to collect image data of a target area at a preset frequency; Inputting the collected image data into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, wherein the judgment result indicates whether there is a strip steel in a target position in the target area; When the judgment result is that the strip steel does not exist at the target position, a reminder signal is generated indicating that the strip steel has exited.

2. The method according to claim 1, characterized in that After obtaining the judgment result of the target residual network model output, the method further includes: When the judgment result is that there is a steel strip at the target position, the process returns to the step of controlling the imaging device to collect image data of the target area at a preset frequency until the judgment result is that there is no steel strip at the target position.

3. The method according to claim 1, characterized in that Before inputting the collected image data into the pre-trained target residual network model, the method further includes: Construct the initial residual network model based on the convolutional layer, maximum pooling layer, attention mechanism layer, average pooling layer and fully connected layer; The initial residual network model is trained to obtain the target residual network model.

4. The method according to claim 3, characterized in that The initial residual network model is constructed according to the convolution layer, the maximum pooling layer, the attention mechanism layer, the average pooling layer and the full connection layer, including: Constructing a first component block according to the convolution layer and the maximum pooling layer; Constructing a second component block according to the convolutional layer and the attention mechanism layer; Constructing a third component block according to the attention mechanism layer, the average pooling layer and the fully connected layer; The initial residual network model is constructed according to the first component block, the third component block and at least four groups of the second component blocks, each group of the second component blocks includes at least one second component block.

5. The method according to claim 4, characterized in that When there are 4 groups of the second component blocks, the number of the second component blocks contained in each group of the second component blocks is 3, 4, 6, and 3 respectively.

6. The method according to claim 4, characterized in that The constructing the initial residual network model according to the first component block, the third component block and at least four groups of the second component blocks includes: The first component block, at least four groups of the second component blocks, and the third component block are connected in sequence to construct the initial residual network model.

7. The method according to any one of claims 3 to 6, characterized in that: The attention mechanism layer is an ECA attention mechanism layer.

8. A strip steel exit detection device, characterized in that: The device comprises: An image acquisition module, used to control the image device to acquire image data of the target area at a preset frequency in response to a strip steel detection start signal; A judgment module, used for inputting the collected image data into a pre-trained target residual network model to obtain a judgment result output by the target residual network model, wherein the judgment result indicates whether there is a strip steel at a target position in the target area; The exit module is used to generate a reminder signal that the strip steel has exited when the judgment result is that the strip steel does not exist at the target position.

9. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute to implement a strip exit detection method as claimed in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, when the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to implement a strip exit detection method as described in any one of claims 1 to 7.

Images