Slicing machine broken material detection method and device, computer equipment and readable storage medium

Abstract

The invention relates to a slicing machine broken material detection method and device, computer equipment and a readable storage medium. The method comprises the following steps: acquiring material transportation image data of a to-be-detected slicing machine at the current moment, inputting a material transportation image into a pre-constructed material breakage detection model, extracting material transportation image local features corresponding to the material transportation image data through the material breakage detection model, and finally, according to the material transportation image local features, carrying out material breakage detection on the to-be-detected slicing machine. And obtaining a material breakage detection result of the slicing machine to be detected at the current moment. The material transportation image data of the to-be-detected slicing machine at the current moment are collected, and the material transportation image is input into the material breakage detection model for material breakage detection, so that the material breakage detection result of the to-be-detected slicing machine at the current moment is obtained, and the problem of low material breakage detection accuracy caused by various conditions existing in operation of the slicing machine is solved; the detection accuracy and the detection efficiency of material breakage detection are improved.

Description

Technical Field

[0001] The present application relates to the technical field of cigarette cut tobacco processing, and particularly to a method and device for detecting material interruption of a slicing machine, a computer device, a computer-readable storage medium, and a computer program product. Background Art

[0002] In the current production process of the cut tobacco production line in a cigarette factory, as an important device, the operation state of the slicing machine plays a decisive role in formulating the production plan. At present, there is a phenomenon of material interruption in the operation of the slicing machine due to various reasons. The existing material interruption detection of the slicing machine usually uses a diffuse reflection switch to detect the material interruption of the conveyor belt of the sliced tobacco blocks.

[0003] However, the current method of using a diffuse reflection switch to detect material interruption has the problem of low detection accuracy. Summary of the Invention

[0004] Based on this, in view of the above technical problems, it is necessary to provide a method and device for detecting material interruption of a slicing machine, a computer device, a computer-readable storage medium, and a computer program product that can improve the detection accuracy.

[0005] In a first aspect, the present application provides a method for detecting material interruption of a slicing machine, including:

[0006] Obtaining the material transportation image data of the slicing machine to be detected at the current moment;

[0007] Inputting the material transportation image data into a pre-constructed material interruption detection model, and extracting the local features of the material transportation image corresponding to the material transportation image data through the material interruption detection model;

[0008] Obtaining the material interruption detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image.

[0009] In one of the embodiments, the material interruption detection model includes a plurality of local feature extraction units;

[0010] Inputting the material transportation image data into a pre-constructed material interruption detection model, and extracting the local features of the material transportation image corresponding to the material transportation image data through the material interruption detection model, including:

[0011] Inputting the material transportation image data into the current local feature extraction unit, and obtaining the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit; the current local feature extraction unit is any one of the plurality of local feature extraction units;

[0012] When the current local feature extraction unit is not the last one among multiple local feature extraction units, use the next local feature extraction unit of the current local feature extraction unit as the new current local feature extraction unit, and return to execute the step of obtaining the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit until the current local feature extraction unit is the last one among multiple local feature extraction units, and use the current local features of the material transportation image as the initial local features of the material transportation image;

[0013] Flatten the initial local features of the material transportation image to obtain the local features of the material transportation image.

[0014] In one embodiment, the current local feature extraction unit includes a convolutional layer and a max pooling layer;

[0015] Input the material transportation image data into the current local feature extraction unit to obtain the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit, including:

[0016] Input the material transportation image into the convolutional layer for local feature extraction to obtain the current original local features of the material transportation image;

[0017] Input the original local features of the material transportation image into the max pooling layer to reduce the image size and obtain the current local features of the material transportation image.

[0018] In one embodiment, the material shortage detection model further includes multiple fully connected layers;

[0019] Obtain the material shortage detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image, including:

[0020] Input the local features of the material transportation image into the current fully connected layer, and obtain the current material shortage detection result of the slicing machine to be detected at the current moment according to the current weight matrix, the current activation function associated with the current fully connected layer, and the local features of the material transportation image;

[0021] When the current fully connected layer is not the last one among multiple fully connected layers, use the next fully connected layer of the current fully connected layer as the new current fully connected layer, and return to execute the step of inputting the local features of the material transportation image into the current fully connected layer to obtain the current material shortage detection result of the slicing machine to be detected at the current moment until the current fully connected layer is the last one among multiple fully connected layers, and use the current material shortage detection result as the material shortage detection result of the slicing machine to be detected at the current moment.

[0022] In an exemplary embodiment, obtaining the material transportation image data of the slicing machine to be detected at the current moment includes:

[0023] Collect the original material transportation image data of the slicer to be detected at the current moment through the on-site camera;

[0024] Remove Gaussian noise and salt-and-pepper noise from the original material transportation image data to obtain denoised material transportation image data;

[0025] Enhance the contrast and sharpness of the denoised material transportation image data to obtain enhanced material transportation image data;

[0026] Crop the image size of the enhanced material transportation image data to obtain material transportation image data.

[0027] In one embodiment, after obtaining the material shortage detection result of the slicer to be detected at the current moment, the method further includes:

[0028] Display the material transportation image data and the material shortage detection result in a preset user interface;

[0029] And / or

[0030] When the material shortage detection result indicates that the slicer to be detected has a material shortage, generate a material shortage signal and send it to the section control PLC; the material shortage signal is used to instruct the section control PLC to drive a preset alarm light and buzzer for material shortage alarm.

[0031] In a second aspect, the present application also provides a slicer material shortage detection device, including:

[0032] A data acquisition module for acquiring the material transportation image data of the slicer to be detected at the current moment;

[0033] A feature extraction module for inputting the material transportation image data into a pre-constructed material shortage detection model, and extracting the local material transportation image features corresponding to the material transportation image data through the material shortage detection model;

[0034] A material shortage detection module for obtaining the material shortage detection result of the slicer to be detected at the current moment according to the local material transportation image features.

[0035] In a third aspect, the present application also provides a computer device, including a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0036] Obtain the material transportation image data of the slicer to be detected at the current moment;

[0037] Input the material transportation image data into a pre-constructed material shortage detection model, and extract the local material transportation image features corresponding to the material transportation image data through the material shortage detection model;

[0038] Based on the local features of the material transportation image, obtain the material breakage detection result of the slicing machine to be detected at the current moment.

[0039] Fourthly, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0040] Obtain the material transportation image data of the slicing machine to be detected at the current moment;

[0041] Input the material transportation image data into a pre-constructed material breakage detection model, and extract the local features of the material transportation image corresponding to the material transportation image data through the material breakage detection model;

[0042] Based on the local features of the material transportation image, obtain the material breakage detection result of the slicing machine to be detected at the current moment.

[0043] Fifthly, the present application further provides a computer program product, including a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0044] Obtain the material transportation image data of the slicing machine to be detected at the current moment;

[0045] Input the material transportation image data into a pre-constructed material breakage detection model, and extract the local features of the material transportation image corresponding to the material transportation image data through the material breakage detection model;

[0046] Based on the local features of the material transportation image, obtain the material breakage detection result of the slicing machine to be detected at the current moment.

[0047] For the above-mentioned material breakage detection method, device, computer device, computer-readable storage medium and computer program product of the slicing machine, by obtaining the material transportation image data of the slicing machine to be detected at the current moment, inputting the material transportation image into a pre-constructed material breakage detection model, extracting the local features of the material transportation image corresponding to the material transportation image data through the material breakage detection model, and finally obtaining the material breakage detection result of the slicing machine to be detected at the current moment based on the local features of the material transportation image. By collecting the material transportation image data of the slicing machine to be detected at the current moment and inputting the material transportation image into the material breakage detection model for material breakage detection, the material breakage detection result of the slicing machine to be detected at the current moment is obtained, avoiding the problem of low accuracy of material breakage detection caused by various situations existing in the operation of the slicing machine, and realizing the improvement of the detection accuracy and detection efficiency of material breakage detection. Description of the Drawings

[0048] To more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the accompanying drawings required for the description of the embodiments of the present application or related technologies. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related accompanying drawings can also be obtained based on these drawings.

[0049] Figure 1 It is an application environment diagram of the slicer stock breakage detection method in an embodiment;

[0050] Figure 2 It is a flowchart of the slicer stock breakage detection method in an embodiment;

[0051] Figure 3 It is a flowchart of the slicer stock breakage detection method in another embodiment;

[0052] Figure 4 It is an overall framework result diagram of the slicer stock breakage detection method in an embodiment;

[0053] Figure 5 It is a flowchart of the image acquisition method in an embodiment;

[0054] Figure 6 It is an interface display diagram of the user interface in another embodiment;

[0055] Figure 7 It is a structural block diagram of the slicer stock breakage detection device in an embodiment;

[0056] Figure 8 It is an internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0057] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0058] The slicer stock breakage detection method provided by the embodiments of the present application can be applied to, for example Figure 1In the application environment shown. Among them, the camera installed at the tobacco material transportation site communicates with the server 102 through the network. The tobacco material transportation site includes equipment such as a material transportation conveyor belt, multiple slicing machines, and industrial high-definition cameras. The data storage system can store the data that the server 102 needs to process. The data storage system can be integrated on the server 102, or placed in the cloud or other network servers. The server 104 obtains the material transportation image data of the slicing machine to be detected at the current moment from the camera installed at the tobacco material transportation site, inputs the material transportation image data into a pre-constructed stock break detection model, extracts the local features of the material transportation image corresponding to the material transportation image data through the stock break detection model, and finally obtains the stock break detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image. Among them, the server 102 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0059] In an exemplary embodiment, as Figure 2 shown, a method for detecting stock break of a slicing machine is provided. Taking the server 102 in Figure 1 as an example, the method includes the following steps S201 to S203. Among them:

[0060] Step S201, obtain the material transportation image data of the slicing machine to be detected at the current moment.

[0061] Among them, the slicing machine can be understood as a device used to cut tobacco leaves into uniform thin slices, usually used in the tobacco processing and production process; the material transportation image data can be understood as multiple photos of the materials required for preparing tobacco being transported on the conveyor belt before entering the slicing machine.

[0062] Optionally, the server 102 obtains the material transportation image data of the slicing machine to be detected at the current moment collected by the camera installed at the tobacco transportation site before entering the slicing machine, and simultaneously performs real-time data preprocessing on the collected material transportation image data. A series of data signal processing algorithms can be used to remove the noise in the image, and the contrast and clarity of the image can be enhanced through image enhancement technology. The size of the processed image data is cropped, and then the acquisition of the material transportation image data is considered completed. By performing real-time processing on the collected image data, the effectiveness and availability of the image data are improved, thereby improving the detection accuracy of the stock break detection, and cropping the size before inputting the model speeds up the detection speed of the stock break detection.

[0063] Step S202, input the material transportation image data into a pre-constructed stock break detection model, and extract the local features of the material transportation image corresponding to the material transportation image data through the stock break detection model.

[0064] Among them, the stock-breaking detection model can be understood as a model used to judge whether there is a risk of stock-breaking in the slicing machine to be detected according to the material transportation image data. The output of the model can be the classification result of the image data or the probability values of having material and stock-breaking. The local features of the material transportation image can be understood as subtle features such as the edges and textures of the material.

[0065] Exemplarily, the server 102 inputs the material transportation image data into a pre-constructed stock-breaking detection model, and uses the convolutional layer in the stock-breaking detection model to perform convolutional operations and flattening processing on the material transportation image data, so as to extract the local features of the material transportation image corresponding to the material transportation image data. Through convolutional operations, features that are more expressive for the current situation can be extracted more deeply, and flattening processing is performed to reduce the data dimension, laying a data foundation for subsequent stock-breaking detection.

[0066] Step S203, according to the local features of the material transportation image, obtain the stock-breaking detection result of the slicing machine to be detected at the current moment.

[0067] Optionally, the server 102 inputs the local features of the material transportation image into the fully connected layer in the stock-breaking detection model for detection. The fully connected layer will comprehensively integrate all the feature information extracted before, and through the operation of the weight matrix, map it to a vector space with a fixed dimension. In this process, the model will highly abstract and integrate the features of stock-breaking and non-stock-breaking. For example, the model will learn the different distribution patterns of stock-breaking images and normal images in the feature vector space, so as to judge whether the current image is stock-breaking or non-stock-breaking. The number of output nodes of the last fully connected layer corresponds to the classification categories. The fully connected layer comprehensively processes all the local features extracted by the previous convolutional layers, ensuring that the model can use all important information for decision-making. This integration helps to capture the global context of the image features, thereby improving the credibility of the stock-breaking detection result.

[0068] In the above slicing machine stock-breaking detection method, by obtaining the material transportation image data of the slicing machine to be detected at the current moment, inputting the material transportation image into a pre-constructed stock-breaking detection model, extracting the local features of the material transportation image corresponding to the material transportation image data through the stock-breaking detection model, and finally obtaining the stock-breaking detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image. By collecting the material transportation image data of the slicing machine to be detected at the current moment and inputting the material transportation image into the stock-breaking detection model for stock-breaking detection, the stock-breaking detection result of the slicing machine to be detected at the current moment is obtained, avoiding the problem of low accuracy of stock-breaking detection caused by various situations in the operation of the slicing machine, and realizing the improvement of the detection accuracy and detection efficiency of stock-breaking detection.

[0069] In one embodiment, the stock-breaking detection model includes multiple local feature extraction units;

[0070] Input the material transportation image data into the pre-constructed stock-breaking detection model, and extract the local features of the material transportation image corresponding to the material transportation image data through the stock-breaking detection model, including:

[0071] Input the material transportation image data into the current local feature extraction unit, and obtain the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit; the current local feature extraction unit is any one of the multiple local feature extraction units;

[0072] In the case that the current local feature extraction unit is not the last one of the multiple local feature extraction units, take the next local feature extraction unit of the current local feature extraction unit as the new current local feature extraction unit, and return to execute the step of obtaining the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit until the current local feature extraction unit is the last one of the multiple local feature extraction units, and take the current local features of the material transportation image as the initial local features of the material transportation image;

[0073] Perform flattening processing on the initial local features of the material transportation image to obtain the local features of the material transportation image.

[0074] Among them, the flattening processing can be understood as data dimensionality reduction processing, which transforms a multi-dimensional array into a one-dimensional vector.

[0075] Exemplarily, the server 102 inputs the material transportation image data into the current local feature extraction unit, and obtains the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit. In the case that the current local feature extraction unit is not the last one of the multiple local feature extraction units, take the next local feature extraction unit of the current local feature extraction unit as the new current local feature extraction unit, and return to execute the step of obtaining the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit until the current local feature extraction unit is the last one of the multiple local feature extraction units, and take the current local features of the material transportation image as the initial local features of the material transportation image; the multiple local feature extraction units in the stock-breaking detection model are arranged in sequence, and the output of the previous local feature extraction unit is used as the input of the next local feature extraction unit. Through the feature extraction of multiple units, the local features are continuously abstracted and strengthened, thereby enhancing the representation ability of the initial local features of the material transportation image.

[0076] Subsequently, the local features of the initial material transportation image are transformed from a multi-dimensional array into a one-dimensional vector of the local features of the material transportation image. Through multi-layer feature extraction and integration, the model can better adapt to inputs under different conditions. The features extracted layer by layer have greater generalization ability, enabling the model to handle unseen data and provide stable prediction performance. The flattening process can greatly reduce the data processing volume and speed up the prediction speed.

[0077] In one embodiment, the current local feature extraction unit includes a convolutional layer and a max-pooling layer;

[0078] Input the material transportation image data into the current local feature extraction unit, and obtain the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit, including:

[0079] Input the material transportation image into the convolutional layer for local feature extraction to obtain the current original local features of the material transportation image;

[0080] Input the original local features of the material transportation image into the max-pooling layer to reduce the image size and obtain the current local features of the material transportation image.

[0081] Optionally, input the material transportation image data into the convolutional layer in the current local feature extraction unit for local feature extraction to obtain the current original local features of the material transportation image, and input the original local features of the material transportation image into the max-pooling layer in the current local feature extraction unit to reduce the image size and obtain the current local features of the material transportation image.

[0082] The entire VGG16 (Visual Geometry Group 16-layer, a deep convolutional neural network architecture), which is the aforementioned blanking detection model, contains 13 convolutional layers. Taking the first convolutional layer (block1_conv1) as an example, its kernel size is (3,3) and the number of filters is 64. In the cases of blanking and non-blanking, this layer performs convolutional operations on the input image. For the input image data, the convolutional kernel slides on the image, and local features of the image are extracted through convolutional calculations. For example, for an image area where blanking may occur, the convolutional layer can capture subtle feature changes such as the edges and textures of the material; for an image without blanking, it will also extract the feature information of the material in its normal state. These features are continuously strengthened and abstracted as the convolutional layers progress. After being processed by the block1_conv1 layer, the feature information of the image is converted into 64 feature maps, and the size remains 224*224. Subsequent convolutional layers such as block1_conv2, etc., will continue to perform convolutional operations based on the output of the previous layer, continuously extracting higher-level and more abstract features, and the number and size of the feature maps will also change accordingly according to the convolutional kernel and pooling operations. For example, after passing through the block1_pool layer (max pooling layer), the image size becomes 112*112. After a series of convolutional layers and pooling layers, the features of the image are fully extracted and compressed. Before entering the fully connected layer, the feature maps after multiple convolutions and poolings are flattened. Through the successive processing of convolution and max pooling, the model can better extract and fuse features, thus significantly improving the accuracy of blanking detection. The reduction in the size of the feature maps and the flattening process reduce the computational cost and improve the real-time detection ability.

[0083] In an exemplary embodiment, the blanking detection model further includes a plurality of fully connected layers;

[0084] Based on the local features of the material transportation image, obtain the blanking detection result of the slicing machine to be detected at the current moment, including:

[0085] Input the local features of the material transportation image into the current fully connected layer, and based on the current weight matrix, current activation function associated with the current fully connected layer, and the local features of the material transportation image, obtain the current blanking detection result of the slicing machine to be detected at the current moment;

[0086] In the case where the current fully connected layer is not the last one among the plurality of fully connected layers, take the next fully connected layer of the current fully connected layer as the new current fully connected layer, and return to execute the step of inputting the local features of the material transportation image into the current fully connected layer to obtain the current blanking detection result of the slicing machine to be detected at the current moment, until the current fully connected layer is the last one among the plurality of fully connected layers, and take the current blanking detection result as the blanking detection result of the slicing machine to be detected at the current moment.

[0087] Exemplarily, the VGG16 model has three fully connected layers, and the server 102 inputs the local features of the material transportation image into these fully connected layers. The fully connected layers will synthesize all the feature information extracted previously and map it to a vector space of a fixed dimension through the operation of the weight matrix. In this process, the model will highly abstract and integrate the features of material breakage and non-breakage. For example, the model will learn the different distribution patterns of the material breakage image and the normal image in the feature vector space, so as to determine whether the current image belongs to material breakage or non-breakage. The number of output nodes of the last fully connected layer corresponds to the classification categories (in this system, there are two categories: material breakage and non-breakage).

[0088] After being processed layer by layer by the VGG16 model, the last fully connected layer will output a two-dimensional vector (corresponding to the two categories of material breakage and non-breakage). Each element in the vector represents the score or probability value of the corresponding category (the score is converted into a probability through the softmax function). If the probability value representing the material breakage category exceeds the set threshold of 0.5, the system determines it as material breakage; if it does not exceed the threshold and the probability value representing the non-breakage category is relatively high, it is determined as non-breakage. For example, when the output vector is [0.9, 0.1], it indicates that the model believes the probability that the current image is material breakage is 0.9, and the system determines it as a material breakage situation; if the output vector is [0.2, 0.8], it is determined as non-breakage. By comprehensively and abstracting features, the fully connected layer effectively improves the accuracy of material breakage detection, enabling the model to make accurate judgments in various situations. In addition, by expressing the result in the form of probability, the model not only tells the user the judgment result but also provides a basis for the judgment, facilitating corresponding processing according to different situations.

[0089] In one embodiment, obtaining the material transportation image data of the slicing machine to be detected at the current moment includes:

[0090] Collecting the original material transportation image data of the slicing machine to be detected at the current moment through a camera installed on-site; removing Gaussian noise and salt-and-pepper noise from the original material transportation image data to obtain denoised material transportation image data; enhancing the contrast and clarity of the denoised material transportation image data to obtain enhanced material transportation image data; and cropping the size of the enhanced material transportation image data to obtain the material transportation image data.

[0091] Optionally, an industrial-grade high-definition camera is selected, which can accurately capture the detailed changes in the moment of material operation. The lens uses optical materials with high light transmittance and low distortion. It is ensured that high-contrast and clear-detail images can be obtained under various complex lighting conditions, such as direct strong light and dim light.

[0092] During the image acquisition process, to improve the image quality, the system will perform real-time preprocessing on the acquired images. By adopting a series of digital signal processing algorithms, such as the denoising algorithm based on wavelet transform, various types of noise interferences in the images, such as Gaussian noise and salt-and-pepper noise, can be effectively removed, while the edge and detail information of the images are retained. At the same time, techniques such as histogram equalization and adaptive contrast stretching are used to enhance the contrast and clarity of the images. In the cropping step, according to the model input requirements, the images are automatically and precisely cropped to form images of size 224*224. By denoising the images and enhancing their contrast and clarity, the characteristics of the materials are made more prominent, providing a high-quality data basis for subsequent detection and determination, and cropping the images according to the model input requirements to ensure that the images are more compliant with the requirements of deep learning model processing and analysis.

[0093] In one embodiment, after obtaining the material shortage detection result of the slicing machine to be detected at the current moment, the method further includes:

[0094] In a preset user interaction interface, display the material transportation image data and the material shortage detection result; and / or, in the case where the material shortage detection result indicates that the slicing machine to be detected has a material shortage, generate a material shortage signal and send it to the section control PLC; the material shortage signal is used to instruct the section control PLC to drive a pre-set alarm light and buzzer to give a material shortage alarm.

[0095] Exemplarily, the interface design follows the principle of simple and intuitive user experience, adopts a clear layout and visual elements, enabling users to quickly obtain key information. In the material shortage determination result display area, the current material shortage determination result, such as "with material" or "material shortage", is displayed in a prominent font and color in real time, and the confidence information of the determination is attached. At the same time, a historical material shortage record query function is also provided, and users can view the details of material shortage events in a past period of time through time filtering, including information such as the time when the material shortage occurred and the duration.

[0096] To facilitate users to monitor the production process more deeply, the interaction interface also integrates a real-time image preview function. Users can view the images of the slicing machine materials in real time on the web page, and the image refresh rate is synchronized with the acquisition system to ensure that users can observe the operation of the materials in a timely manner. Through this interaction interface, users can remotely and conveniently monitor the operation of the slicing machine without being on-site in the workshop, realizing real-time control of the production process.

[0097] and / or

[0098] When the detection and determination system recognizes a material shortage situation, the alarm output system responds quickly and transmits the material shortage signal to the section control PLC with extremely high priority and precision. To ensure the reliability of signal transmission, the system adopts a redundant design and a data verification mechanism, which verifies the data multiple times during the signal transmission process. Once data errors or losses are found, retransmission is immediately carried out to ensure that the material shortage signal is accurately sent to the section control PLC.

[0099] After receiving the material shortage signal, the section control PLC immediately drives devices such as alarm lights and buzzers according to the pre-written logic program. The alarm lights are designed with high brightness and eye-catching colors, and the buzzers emit high-decibel sound signals with specific frequencies to enhance the alarm effect.

[0100] In an exemplary embodiment, as Figure 3 shown, a specific implementation of a method for detecting material shortage in a slicing machine is provided. As Figure 4 shown in the framework, it mainly consists of four parts: an image acquisition system, a detection and determination system, an alarm output system, and a display interface (the data appearing below are all specific examples, rather than limiting that the present application can only be implemented in this case). Among them:

[0101] I. Image acquisition system:

[0102] As Figure 5 shown, this system selects an industrial-grade high-definition camera, which can accurately capture the detailed changes of the material during operation. The lens uses optical materials with high light transmittance and low distortion to ensure that high-contrast and detailed images can be obtained under various complex lighting conditions, such as direct strong light and dim light.

[0103] During the image acquisition process, to improve the image quality, the system will perform real-time preprocessing on the acquired images. A series of digital signal processing algorithms are adopted, such as denoising algorithms based on wavelet transform, which can effectively remove various types of noise interference such as Gaussian noise and salt-and-pepper noise in the images, retain the edges and details of the images, and at the same time use techniques such as histogram equalization and adaptive contrast stretching to enhance the contrast and clarity of the images. In the cropping link, according to the requirements of model training, the images are automatically and accurately cropped to form images of 224*224 size. Image enhancement uses techniques such as histogram equalization and adaptive contrast stretching to enhance the contrast and clarity of the images, making the features of the material more prominent, providing a high-quality data basis for subsequent detection and determination, and ensuring that the images can better meet the requirements of deep learning model processing and analysis. During the training image acquisition stage, a total of 6214 pictures were collected.

[0104] II. Detection and determination system:

[0105] The core of the detection and determination system is based on the VGG16 neural network model. The model is a convolutional neural network model with 16 weight layers, including 13 convolutional layers and 3 fully connected layers. The ReLU function is used as the activation function.

[0106] Model training: Before model training, the images collected for training are classified into 3,208 qualified photos and 3,006 unqualified photos. Based on this, model training is carried out. The total number of training times is one hundred, the batch size is 32 each time, and the number of steps for each training is automatically obtained by dividing the total number of training by the batch size. The gradient optimization algorithm for model training uses the Adam optimizer, which combines the advantages of Adagrad (Adaptive Gradient Algorithm) and RMSProp (Root Mean Square Propagation Algorithm). It can adaptively adjust the learning rate according to the gradient history information of each parameter. It estimates the update direction and step size of the parameters by calculating the first-order moment estimate (mean) and second-order moment estimate (variance) of the gradient. And it can converge faster during the training process, continuously improving the accuracy and generalization ability of the model.

[0107] After the training is completed, multiple rounds of strict tests and verifications are carried out to ensure the reliability and accuracy of the model. During the verification process, multiple evaluation metrics such as accuracy, recall rate, F1 value, etc. are used to comprehensively evaluate the model performance.

[0108] Model application: The entire VGG16 model contains 13 convolutional layers. Taking the first convolutional layer (block1_conv1) as an example, the size of its convolutional kernel is (3,3), and the number of filters is 64. In the two cases of unqualified and qualified, this layer will perform convolutional operations on the input image. For the input image data, the convolutional kernel slides on the image, and local features of the image are extracted through convolutional calculation. For example, for the image area where unqualified may exist, the convolutional layer can capture subtle feature changes such as the edges and textures of the material; for the qualified image, the feature information of the material in the normal state will also be extracted. These features will be continuously strengthened and abstracted as the convolutional layers progress. After being processed by the block1_conv1 layer, the feature information of the image is converted into 64 feature maps, and the size remains 224*224. Subsequent convolutional layers such as block1_conv2, etc., will continue to perform convolutional operations based on the output of the previous layer, continuously extracting more advanced and abstract features, and the number and size of the feature maps will also change accordingly according to the convolutional kernel and pooling operations. For example, after passing through the block1_pool layer (max pooling layer), the image size becomes 112*112.

[0109] After a series of convolutional layers and pooling layers, the features of the image are fully extracted and compressed. Before entering the fully connected layers, the feature maps after multiple convolutional and pooling operations are flattened. The VGG16 model has 3 fully connected layers, and these fully connected layers take the flattened feature vectors as input. The fully connected layers synthesize all the feature information extracted previously and, through operations with weight matrices, map it to a vector space of a fixed dimension. In this process, the model highly abstracts and integrates the features of broken and unbroken materials. For example, the model will learn the different distribution patterns of broken material images and normal images in the feature vector space, so as to determine whether the current image belongs to broken or unbroken materials. The number of output nodes in the last fully connected layer corresponds to the classification categories (in this system, there are two categories: broken and unbroken).

[0110] After passing through the various layers of the VGG16 model, the last fully connected layer outputs a two-dimensional vector (corresponding to the two categories of broken and unbroken materials). Each element in the vector represents the score or probability value of the corresponding category (the scores are converted to probabilities through the softmax function). If the probability value representing the broken material category exceeds the set threshold of 0.5, the system determines it as a broken material; if it does not exceed the threshold and the probability value representing the unbroken material category is higher, it is determined as unbroken. For example, when the output vector is [0.9, 0.1], it indicates that the model believes the probability that the current image is a broken material is 0.9, and the system determines it as a broken material situation; if the output vector is [0.2, 0.8], it is determined as unbroken.

[0111] III. Alarm Output System:

[0112] The alarm output system constructs a stable, efficient and highly anti-interference communication link, using the advanced industrial communication protocol OPCUA (Open Platform Communications Unified Architecture) to ensure seamless connection with the section control PLC (Programmable Logic Controller). The OPCUA protocol has the advantages of cross-platform, high security, reliable data transmission, etc., and can operate stably in a complex industrial network environment to ensure the accuracy and real-time of data transmission.

[0113] When the detection and determination system identifies a broken material situation, the alarm output system responds quickly and accurately transmits the broken material signal to the section control PLC with extremely high priority. To ensure the reliability of signal transmission, the system adopts a redundant design and a data verification mechanism, and verifies the data multiple times during the signal transmission process. Once data errors or losses are found, retransmission is immediately carried out to ensure that the broken material signal is accurately delivered to the section control PLC.

[0114] After receiving the material shortage signal, the section control PLC immediately drives devices such as the alarm lamp and buzzer according to the pre-written logic program. The alarm lamp is designed with a high-brightness and eye-catching color, and the buzzer emits a high-decibel sound signal with a specific frequency to enhance the alarm effect.

[0115] IV. Interaction Interface:

[0116] As Figure 6 shown, the interaction interface is developed using cutting-edge web technologies. Based on standard technology frameworks such as HTML5 (HyperText Markup Language), CSS3 (Cascading Style Sheets), and JavaScript (scripting language), combined with the nods.js (runtime environment) framework to achieve overall web front-end and back-end data management.

[0117] The interface design follows the principle of simple and intuitive user experience, adopting a clear layout and visual elements, enabling users to quickly obtain key information. In the area for displaying the material shortage judgment result, the current material shortage judgment result, such as "with material" or "material shortage", is displayed in real time with a prominent font and color, and the confidence information of the judgment is attached. At the same time, a historical material shortage record query function is also provided, allowing users to view the details of material shortage events within a certain period in the past through time filtering, including information such as the time when the material shortage occurred and the duration.

[0118] To facilitate users to conduct more in-depth monitoring of the production process, the interaction interface also integrates a real-time image preview function. Users can view the images of the slicer material in real time on the web page, and the image refresh rate is synchronized with the acquisition system to ensure that users can timely observe the operation of the material. Through this interaction interface, users can remotely and conveniently monitor the operation of the slicer without being on-site in the workshop, achieving real-time control of the production process.

[0119] Compared with the prior art, the present application has the following technical advantages:

[0120] 1. The on-site image data is collected through an industrial-grade high-definition camera, and a series of data processing operations such as denoising, contrast enhancement, and clarity enhancement are performed on the collected image data, and the processed image is cropped in size, improving the effectiveness and usability of the image data, and thus improving the detection accuracy of material shortage detection.

[0121] 2. After repeated optimization and verification, the material shortage detection model shows extremely high accuracy in material shortage detection, and can stably and reliably output the judgment result of whether there is a material shortage, providing accurate decision-making basis for the production process.

[0122] 3. Through this intuitive acoustic and optical method, on-site workers are reminded to handle the material shortage problem in a timely manner, ensuring the continuity of the production process and effectively avoiding problems such as production interruption, material waste, and product quality decline caused by the failure to detect material shortage in a timely manner. At the same time, the alarm output system also has an event recording function, which can record information such as the time and type of each alarm, facilitating subsequent traceability and analysis of abnormal situations in the production process.

[0123] 4. Through the display of relevant information of the slicing machine to be detected on the interaction interface, users can remotely and conveniently monitor the operation of the slicing machine without being on-site in the workshop, achieving real-time control of the production process. This not only improves the efficiency of production management but also reduces the workload of manual inspections, enabling managers to make decisions in a timely manner and enhancing the intelligent management level of the entire production system.

[0124] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are sequentially shown according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0125] Based on the same inventive concept, the embodiments of the present application also provide a slicing machine material shortage detection device for implementing the slicing machine material shortage detection method described above. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the slicing machine material shortage detection device provided below can refer to the limitations on the slicing machine material shortage detection method in the above text and will not be repeated here.

[0126] In an exemplary embodiment, as Figure 7 shown, a slicing machine material shortage detection device is provided, including: a data acquisition module 701, a feature extraction module 702, and a material shortage detection module 703, where:

[0127] The data acquisition module 701 is used to acquire the material transportation image data of the slicing machine to be detected at the current moment;

[0128] The feature extraction module 702 is used to input the material transportation image data into a pre-constructed material shortage detection model, and extract the local features of the material transportation image corresponding to the material transportation image data through the material shortage detection model;

[0129] The blank material detection module 703 is configured to obtain the blank material detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image.

[0130] In one embodiment, the blank material detection model includes a plurality of local feature extraction units; the feature extraction module 702 includes a current feature extraction sub-module, a loop sub-module, and a flattening processing sub-module, wherein:

[0131] The current feature extraction sub-module is configured to input the material transportation image data into the current local feature extraction unit, and obtain the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit; the current local feature extraction unit is any one of the plurality of local feature extraction units;

[0132] The loop sub-module is configured to, when the current local feature extraction unit is not the last one of the plurality of local feature extraction units, use the next local feature extraction unit of the current local feature extraction unit as the new current local feature extraction unit, and return to execute the step of obtaining the current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit, until the current local feature extraction unit is the last one of the plurality of local feature extraction units, and use the current local features of the material transportation image as the initial local features of the material transportation image;

[0133] The flattening processing sub-module is configured to perform flattening processing on the initial local features of the material transportation image to obtain the local features of the material transportation image.

[0134] In one of the embodiments, the current local feature extraction unit includes a convolutional layer and a max pooling layer, and the current feature extraction sub-module is further configured to input the material transportation image into the convolutional layer for local feature extraction to obtain the current original local features of the material transportation image; input the original local features of the material transportation image into the max pooling layer to reduce the image size, and obtain the current local features of the material transportation image.

[0135] In an exemplary embodiment, the blanking detection model further includes a plurality of fully connected layers; the blanking detection module 703 is further configured to input the local features of the material transportation image into the current fully connected layer, and obtain the current blanking detection result of the slicing machine to be detected at the current moment according to the current weight matrix, the current activation function associated with the current fully connected layer, and the local features of the material transportation image; when the current fully connected layer is not the last one of the plurality of fully connected layers, the next fully connected layer of the current fully connected layer is used as the new current fully connected layer, and the step of inputting the local features of the material transportation image into the current fully connected layer to obtain the current blanking detection result of the slicing machine to be detected at the current moment is returned and executed until the current fully connected layer is the last one of the plurality of fully connected layers, and the current blanking detection result is used as the blanking detection result of the slicing machine to be detected at the current moment.

[0136] In one embodiment, the data acquisition module 701 is further configured to collect the original material transportation image data of the slicing machine to be detected at the current moment through a camera installed on site; remove Gaussian noise and salt-and-pepper noise in the original material transportation image data to obtain denoised material transportation image data; enhance the contrast and clarity of the denoised material transportation image data to obtain enhanced material transportation image data; and crop the image size of the enhanced material transportation image data to obtain the material transportation image data.

[0137] In one of the embodiments, the blanking detection device is further configured to display the material transportation image data and the blanking detection result in a preset user interface; and / or, when the blanking detection result indicates that the slicing machine to be detected has a blanking situation, generate a blanking signal and send it to the section control PLC; the blanking signal is used to instruct the section control PLC to drive a pre-set alarm light and buzzer to perform blanking alarm.

[0138] Each module in the above-mentioned slicing machine blanking detection device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0139] In an exemplary embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 8As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store material transportation image data, local features of material transportation images, and data on the result of stock breakage detection. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a method for detecting stock breakage of a slicer.

[0140] Those skilled in the art can understand that Figure 8 the structure shown in the figure is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0141] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, it implements the method for detecting stock breakage of a slicer in the above embodiment.

[0142] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the method for detecting stock breakage of a slicer in the above embodiment.

[0143] In an embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, it implements the method for detecting stock breakage of a slicer in the above embodiment.

[0144] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0145] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0146] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope recorded in the present application.

[0147] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method for detecting material breakage of a slicing machine, characterized in that, The method includes: Obtaining material transportation image data of the slicing machine to be detected at the current moment; Inputting the material transportation image data into a pre-constructed material shortage detection model, and extracting local features of the material transportation image corresponding to the material transportation image data through the material shortage detection model; Obtaining a material shortage detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image.

2. The method according to claim 1, wherein The material shortage detection model includes multiple local feature extraction units; The step of inputting the material transportation image data into a pre-constructed material shortage detection model and extracting local features of the material transportation image corresponding to the material transportation image data through the material shortage detection model includes: Inputting the material transportation image data into the current local feature extraction unit, and obtaining current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit; the current local feature extraction unit is any one of the multiple local feature extraction units; When the current local feature extraction unit is not the last one of the multiple local feature extraction units, taking the next local feature extraction unit of the current local feature extraction unit as the new current local feature extraction unit, and returning to execute the step of obtaining current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit until the current local feature extraction unit is the last one of the multiple local feature extraction units, and taking the current local features of the material transportation image as the initial local features of the material transportation image; Performing flattening processing on the initial local features of the material transportation image to obtain the local features of the material transportation image.

3. The method according to claim 2, wherein The current local feature extraction unit includes a convolutional layer and a max pooling layer; The step of inputting the material transportation image data into the current local feature extraction unit and obtaining current local features of the material transportation image corresponding to the material transportation image data through the current local feature extraction unit includes: Inputting the material transportation image into the convolutional layer for local feature extraction to obtain current original local features of the material transportation image; Inputting the original local features of the material transportation image into the max pooling layer to reduce the image size and obtain the current local features of the material transportation image.

4. The method according to any one of claims 1 to 3, characterized in that The material shortage detection model further includes multiple fully connected layers; The step of obtaining a material shortage detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image includes: Inputting the local features of the material transportation image into the current fully connected layer, and obtaining the current material shortage detection result of the slicing machine to be detected at the current moment according to the current weight matrix, current activation function associated with the current fully connected layer, and the local features of the material transportation image. When the current fully connected layer is not the last one among the multiple fully connected layers, use the next fully connected layer of the current fully connected layer as the new current fully connected layer, and return to execute the step of inputting the local features of the material transportation image into the current fully connected layer to obtain the current material breakage detection result of the slicing machine to be detected at the current moment, until the current fully connected layer is the last one among the multiple fully connected layers, and use the current material breakage detection result as the material breakage detection result of the slicing machine to be detected at the current moment.

5. The method according to claim 1, wherein The obtaining of the material transportation image data of the slicing machine to be detected at the current moment includes: Collect the original material transportation image data of the slicing machine to be detected at the current moment through a camera installed on site; Remove Gaussian noise and salt-and-pepper noise from the original material transportation image data to obtain denoised material transportation image data; Enhance the contrast and clarity of the denoised material transportation image data to obtain enhanced material transportation image data; Crop the size of the enhanced material transportation image data to obtain the material transportation image data.

6. The method according to claim 1, wherein After obtaining the material breakage detection result of the slicing machine to be detected at the current moment, the method further includes: Display the material transportation image data and the material breakage detection result in a preset user interface; and / or When the material breakage detection result indicates that the slicing machine to be detected has a material breakage, generate a material breakage signal and send it to the section control PLC; the material breakage signal is used to instruct the section control PLC to drive a pre-set alarm light and buzzer to perform a material breakage alarm.

7. A material cutting detection device for a slicing machine, characterized in that, The device includes: A data acquisition module for acquiring the material transportation image data of the slicing machine to be detected at the current moment; A feature extraction module for inputting the material transportation image data into a pre-constructed material breakage detection model, and extracting the local features of the material transportation image corresponding to the material transportation image data through the material breakage detection model; A material breakage detection module for obtaining the material breakage detection result of the slicing machine to be detected at the current moment according to the local features of the material transportation image.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 6.

Images