Traceable Internet of Things system based on intelligent recovery

Abstract

The invention discloses a traceable Internet of Things system based on intelligent recovery. The method comprises the following steps: firstly installing an intelligent label on a waste electronic product, then storing and processing product related data of the waste electronic product recorded by the intelligent label so as to carry out recovery value analysis on the waste electronic product, then providing a recovery service of the waste electronic product, and scanning the intelligent label so as to carry out recovery value analysis on the waste electronic product. The method comprises the following steps: acquiring product related data of the waste electronic products according to the received waste electronic products, classifying, packaging and transporting the waste electronic products, finally, disassembling, separating and recycling the recycled waste electronic products, and updating the product related data of the waste electronic products by scanning the intelligent labels. And then the updated product related data of the waste electronic product is fed back to the cloud platform processing module. Therefore, recording and tracing of related data of the product can be realized.

Description

Technical Field

[0001] This application relates to the field of smart recycling, and more specifically, to an Internet of Things system based on smart recycling traceability. Background Technology

[0002] With the widespread use of electronic products and their rapid pace of replacement, the disposal and recycling of waste electronic products has become a significant environmental and resource management issue. Waste electronic products refer to electronic products that have lost their usability or have been discarded by consumers, such as mobile phones, computers, and televisions. Waste electronic products not only consume a large amount of resources but also contain many harmful substances. If not properly recycled and disposed of, they will cause serious harm to the environment and human health.

[0003] Currently, the recycling and processing of waste electronic products mainly relies on manual sorting, dismantling, separation, and reuse. This method is not only inefficient, but also cannot accurately assess the recycling value of waste electronic products and is difficult to trace the source of the products. This can lead to problems such as misjudgment, damage, and pollution in the recycling process.

[0004] Therefore, there is a need for an IoT system based on intelligent recycling and traceability to improve the efficiency and quality of waste electronic products recycling, and reduce resource waste and environmental pollution. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. This application provides an IoT system based on intelligent recycling and traceability, which can realize the recording and traceability of product-related data.

[0006] According to one aspect of this application, an IoT system based on intelligent recycling traceability is provided, comprising:

[0007] A smart tag installation module is used to install smart tags on waste electronic products, wherein the smart tags are used to record product-related data of the waste electronic products;

[0008] The cloud platform processing module is used to store and process the product-related data of the waste electronic products recorded by the smart tags, so as to perform recycling value analysis on the waste electronic products;

[0009] The recycling station module is used to provide recycling services for the waste electronic products, and to obtain product-related data of the waste electronic products by scanning the smart tags, so as to classify, pack and transport the waste electronic products; and

[0010] The waste electronic product processing module is used to dismantle, separate, and reuse the recycled waste electronic products, and update the product-related data of the waste electronic products by scanning the smart tags, and then feed the updated product-related data of the waste electronic products back to the cloud platform processing module.

[0011] Compared to existing technologies, the IoT system based on intelligent recycling and traceability provided in this application first installs smart tags on waste electronic products. Next, it stores and processes product-related data recorded by the smart tags to analyze the recycling value of the waste electronic products. Then, it provides recycling services for the waste electronic products, and by scanning the smart tags, it obtains the product-related data to classify, package, and transport the waste electronic products. Finally, it disassembles, separates, and reuses the recycled waste electronic products, and by scanning the smart tags, it updates the product-related data of the waste electronic products, then feeds the updated product-related data back to the cloud platform processing module. This enables the recording and traceability of product-related data. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. The following drawings are not deliberately drawn to scale according to the actual size, but are intended to show the main idea of ​​this application.

[0013] Figure 1 This is a block diagram of an IoT system based on intelligent recycling and traceability according to an embodiment of this application.

[0014] Figure 2 This is a block diagram of the cloud platform processing module in an IoT system based on intelligent recycling traceability according to an embodiment of this application.

[0015] Figure 3 This is a flowchart of an IoT method based on intelligent recycling traceability according to an embodiment of this application.

[0016] Figure 4 This is an application scenario diagram of an IoT system based on intelligent recycling and traceability according to an embodiment of this application. Detailed Implementation

[0017] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are also within the scope of protection of this application.

[0018] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0019] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0020] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously, as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0021] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0022] To address the aforementioned technical problems, this application proposes an IoT system based on intelligent recycling and traceability. Figure 1 This is a block diagram of an IoT system based on intelligent recycling and traceability according to an embodiment of this application. Figure 1As shown, the IoT system 100 based on intelligent recycling and traceability according to an embodiment of this application includes: a smart tag installation module 110, used to install smart tags on waste electronic products, the smart tags being used to record product-related data of the waste electronic products; a cloud platform processing module 120, used to store and process the product-related data of the waste electronic products recorded by the smart tags, in order to perform recycling value analysis on the waste electronic products; a recycling station module 130, used to provide recycling services for the waste electronic products, and to obtain the product-related data of the waste electronic products by scanning the smart tags, so as to classify, package, and transport the waste electronic products; and a waste electronic product processing module 140, used to disassemble, separate, and reuse the recycled waste electronic products, and to update the product-related data of the waste electronic products by scanning the smart tags, and then feed the updated product-related data of the waste electronic products back to the cloud platform processing module. It should be understood that the development of IoT technology has provided a new solution for the recycling of waste electronic products. By installing smart tags on waste electronic products, the recording and traceability of product-related data can be achieved. Smart tags can record basic information about electronic products, such as model number, production date, and usage status. They can also record images of electronic products to provide more comprehensive information.

[0023] Accordingly, in the aforementioned IoT system based on intelligent recycling and traceability, analyzing the product-related data of the waste electronic products recorded by the smart tags in the cloud platform processing module is crucial for determining the recycling value of these products. This helps decision-makers and recycling stations determine the processing methods and priorities for waste electronic products. Based on this, the technical concept of this application involves acquiring the basic information and images of electronic products from the product-related data provided by the smart tags, and then introducing image processing and semantic understanding algorithms in the backend to analyze the basic information and images of the electronic products. This, combined with the semantic information of both, comprehensively assesses and judges the recycling value level of the waste electronic products. This improves the efficiency and accuracy of waste electronic product recycling, thereby achieving effective resource utilization and sustainable environmental development.

[0024] Figure 2 This is a block diagram of the cloud platform processing module in an IoT system based on intelligent recycling traceability according to an embodiment of this application. Figure 2As shown, the cloud platform processing module 120 includes: a product-related data acquisition unit 121, used to acquire product-related data provided by smart tags installed on the waste electronic products, wherein the product-related data includes basic information of the electronic products and images of the electronic products; an electronic product surface state feature extraction unit 122, used to extract features from the electronic product images using an electronic product surface state feature extractor based on a deep neural network model to obtain an electronic product surface state feature map; an electronic product surface state feature enhancement unit 123, used to perform adaptive attention enhancement processing on the electronic product surface state feature map to obtain an adaptive attention enhanced electronic product surface state feature map; an electronic product basic information semantic encoding unit 124, used to perform semantic encoding on the electronic product basic information to obtain an electronic product basic information semantic encoding feature vector; a cross-modal feature fusion unit 125, used to perform cross-modal feature fusion on the adaptive attention enhanced electronic product surface state feature map and the electronic product basic information semantic encoding feature vector to obtain multimodal electronic product recycling value characterization features; and a recycling value detection unit 126, used to determine the recycling value level label based on the multimodal electronic product recycling value characterization features.

[0025] Specifically, in the technical solution of this application, firstly, product-related data provided by smart tags installed on discarded electronic products is acquired, wherein the product-related data includes basic information of the electronic products and images of the electronic products. Next, an electronic product surface state feature extractor based on a convolutional neural network model, which has excellent performance in extracting latent features from images, is used to perform feature mining on the electronic product images to extract the surface state feature distribution information of the discarded electronic products in the electronic product images, thereby obtaining an electronic product surface state feature map.

[0026] Accordingly, in the electronic product surface state feature extraction unit 122, the deep neural network model is a convolutional neural network model, that is, the electronic product surface state feature extractor based on the deep neural network model is an electronic product surface state feature extractor based on the convolutional neural network model.

[0027] It's worth noting that a Convolutional Neural Network (CNN) is a deep learning model primarily used for image recognition and processing tasks. A CNN consists of multiple layers, including convolutional layers, pooling layers, and fully connected layers. The convolutional layer is the core of the CNN, applying a series of convolutional kernels (also called filters) to the input image to extract local features. Convolutional operations can capture low-level features such as edges, textures, and shapes in an image. Pooling layers reduce the spatial size of the feature map, decreasing the number of parameters while preserving important feature information. Common pooling operations include max pooling and average pooling, which select the maximum or average value within the pooling window as the output, respectively. The fully connected layer transforms the feature maps from the previous layers into the final classification or regression result. In a fully connected layer, each neuron is connected to all neurons in the previous layer, learning weights and biases to combine and classify features. Convolutional neural networks extract high-level features from images through multiple layers of convolution and pooling operations, enabling the network to automatically learn abstract features in images and thus achieve accurate classification or regression prediction of images.

[0028] Furthermore, considering that each channel of the electronic product surface state feature map represents different features related to the surface state of the waste electronic product, some of which are important for judging the recycling value of the waste electronic product, while others are irrelevant interference features, in order to focus attention on important channel features and reduce the influence of irrelevant features on recycling value detection, thereby improving the classifier's ability to identify and judge the recycling value level of waste electronic products, the technical solution of this application further uses an attention mechanism to weight important information in the feature map to improve the perception and recognition ability of key features. Specifically, in the technical solution of this application, the electronic product surface state feature map is further processed through an adaptive attention module to obtain an adaptive attention-enhanced electronic product surface state feature map. It should be understood that the adaptive attention module uses a meta-weight generator to transform the feature map of each channel into a weight value. These weight values ​​are multiplied with the electronic product surface state feature map channel by channel, so that each channel in the feature map receives different degrees of attention, thereby highlighting important channel feature information. In this way, attention can be focused on important features related to the recycling value level detection of waste electronic products, reducing the influence of irrelevant features, thereby improving the classifier's recognition and classification capabilities.

[0029] Accordingly, the electronic product surface state feature enhancement unit 123 is used to: pass the electronic product surface state feature map through an adaptive attention module to obtain the adaptive attention enhanced electronic product surface state feature map.

[0030] Specifically, in one example, the electronic product surface state feature enhancement unit 123 is used to: process the electronic product surface state feature map through the adaptive attention module using the following adaptive enhancement formula to obtain the adaptive attention-enhanced electronic product surface state feature map; wherein, the adaptive enhancement formula is:

[0031]

[0032]

[0033]

[0034] v = pool(F)

[0035] Where F is the surface state feature map of the electronic product, pool(·) represents global mean pooling of each feature matrix along the channel dimension in the feature map, v is the channel feature vector of the surface state feature map of the electronic product, and W a and B a Here, σ represents the weights and biases of the convolutional layer, σ is the activation function, and A is the convolutional feature vector of the channel feature vector. i A is the feature value at each position in the convolutional feature vector. ′ It is a weight vector, ⊙ is the positional dot product, F ′ It is the surface state feature map of the adaptive attention-enhanced electronic product.

[0036] Then, considering that the basic information of the electronic products contains some important characteristics about the waste electronic products, such as model, production date, and usage status, this application's technical solution requires semantic encoding of the basic information of the electronic products to obtain a semantically encoded feature vector. By semantically encoding the basic information of the electronic products, important semantic features such as model, production date, and usage status can be extracted, which helps to further improve the accuracy of assessing and judging the recycling value of waste electronic products.

[0037] It should be understood that the adaptive attention-enhanced electronic product surface state feature map and the electronic product basic information semantic encoding feature vector are semantic feature information about different modalities of the waste electronic product extracted from the electronic product image and the electronic product basic information, respectively capturing the surface state visual feature information and important semantic feature information of the waste electronic product. Therefore, in order to better describe and represent the recycling value of the waste electronic product and improve the accuracy of recycling value assessment and judgment, in the technical solution of this application, a meta-network-based cross-modal feature fusion processor is further used to process the adaptive attention-enhanced electronic product surface state feature map and the electronic product basic information semantic encoding feature vector to obtain a multi-modal electronic product recycling value representation feature map. By using the meta-network-based cross-modal feature fusion processor, the surface state visual features and important semantic features of the waste electronic product can be mapped to the same feature space and weighted fusion is performed channel by channel, so that the feature information of different modalities of the waste electronic product can complement and enhance each other, thereby improving the richness and accuracy of the recycling value representation of the waste electronic product.

[0038] Accordingly, the cross-modal feature fusion unit 125 is used to: process the adaptive attention-enhanced electronic product surface state feature map and the electronic product basic information semantic encoding feature vector using a meta-network-based cross-modal feature fusion unit to obtain a multimodal electronic product recycling value representation feature map as the multimodal electronic product recycling value representation feature.

[0039] Specifically, in one example, the cross-modal feature fusion unit 125 includes: a first convolution subunit for passing the semantic encoding feature vector of the basic information of the electronic product through a point convolutional layer to obtain a first convolutional feature vector; a first correction subunit for passing the first convolutional feature vector through a correction linear unit based on the ReLU function to obtain a first corrected convolutional feature vector; a second convolution subunit for passing the first corrected convolutional feature vector through a point convolutional layer to obtain a second convolutional feature vector; a second correction subunit for passing the second convolutional feature vector through a correction linear unit based on the Sigmoid function to obtain a second corrected convolutional feature vector; and a fusion subunit for fusing the second corrected convolutional feature vector with the adaptive attention-enhanced electronic product surface state feature map to obtain the multimodal electronic product recycling value representation feature map.

[0040] Subsequently, the multimodal electronic product recycling value characterization feature map is processed by a classifier to obtain a classification result, which is used to represent the recycling value level label. In other words, the multimodal recycling value characterization feature information of the waste electronic products is used for classification processing to detect the recycling value of the electronic products. Specifically, in the technical solution of this application, the label of the classifier is a recycling value level label. Therefore, after obtaining the classification result, the recycling value level of the waste electronic products can be assessed and judged based on the classification result. This improves the efficiency and accuracy of waste electronic product recycling, thereby achieving effective resource utilization and sustainable environmental development.

[0041] Accordingly, the recycling value detection unit 126 is used to: pass the multimodal electronic product recycling value characterization feature map through a classifier to obtain a classification result, the classification result being used to represent the recycling value level label.

[0042] As you can understand, the role of a classifier is to learn classification rules and classifiers using given categories and known training data, and then classify (or predict) unknown data. Logistic regression and SVM are commonly used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are needed to form the multi-class classification. However, this is prone to errors and inefficient. A commonly used multi-class classification method is the Softmax classification function.

[0043] Furthermore, in the technical solution of this application, the IoT system based on intelligent recycling and traceability further includes a training module for training the electronic product surface state feature extractor based on a convolutional neural network model, the adaptive attention module, the cross-modal feature fusion unit based on a meta-network, and the classifier.

[0044] In one example, the training module includes: a training data acquisition unit for acquiring training data, the training data including training product-related data provided by smart tags installed on the discarded electronic products, wherein the training product-related data includes basic information of the training electronic products and images of the training electronic products; a training electronic product surface state feature extraction unit for extracting features from the images of the training electronic products using the electronic product surface state feature extractor based on the convolutional neural network model to obtain a training electronic product surface state feature map; a training electronic product surface state feature enhancement unit for passing the training electronic product surface state feature map through the adaptive attention module to obtain a training adaptive attention enhanced electronic product surface state feature map; a training electronic product basic information semantic encoding unit for semantically encoding the basic information of the training electronic products to obtain a training electronic product basic information semantic encoding feature vector; and a training cross-modal... The system includes a feature fusion unit for processing the training adaptive attention-enhanced electronic product surface state feature map and the training electronic product basic information semantic encoding feature vector using the meta-network-based cross-modal feature fusion unit to obtain a training multimodal electronic product recycling value representation feature map; an optimization training unit for performing feature aggregation optimization on the training multimodal electronic product recycling value representation feature vector obtained after expanding the training multimodal electronic product recycling value representation feature map to obtain an optimized training multimodal electronic product recycling value representation feature vector; a classification loss calculation unit for passing the optimized training multimodal electronic product recycling value representation feature vector through the classifier to obtain a classification loss function value; and a loss training unit for training the electronic product surface state feature extractor based on the convolutional neural network model, the adaptive attention module, the meta-network-based cross-modal feature fusion unit, and the classifier using the classification loss function value.

[0045] In the technical solution of this application, the trained electronic product surface state feature map expresses the image semantic features of the trained electronic product image. The various feature matrices follow the channel dimension distribution of the electronic product surface state feature extractor based on the convolutional neural network model. By applying an adaptive attention module to the trained electronic product surface state feature map, different weights are applied to each surface state feature of the trained electronic product through an attention mechanism to make the trained electronic product surface state feature map channel saliency. This does not change the local feature distribution law of the trained adaptive attention-enhanced electronic product surface state feature map. The trained electronic product basic information semantic encoding feature vector represents the contextual semantic association features of the trained electronic product basic information. Therefore, when the cross-modal feature fusion based on meta-networks is used to process the trained adaptive attention-enhanced electronic product surface state feature map and the trained electronic product basic information semantic encoding feature vector, the channel dimension representation of each feature matrix of the trained adaptive attention-enhanced electronic product surface state feature map is constrained by the semantic information of the feature values ​​at each position of the trained electronic product basic information semantic encoding feature vector. This causes the probability density representation of each feature value of the trained multimodal electronic product recycling value representation feature map to become sparse in the probability density domain. In other words, the trained multimodal electronic product recycling value representation feature map suffers from insufficient feature distribution aggregation, which affects the regression convergence effect when classified by a classifier.

[0046] Therefore, the applicant of this application preferably performs feature aggregation optimization on the training multimodal electronic product recycling value representation feature vector, for example denoted as V, when the training multimodal electronic product recycling value representation feature vector is obtained by iterative training of classification and regression through a classifier after each expansion of the training multimodal electronic product recycling value representation feature map.

[0047] Accordingly, in one example, the optimized training unit is further configured to: perform feature aggregation optimization on the training multimodal electronic product recycling value representation feature vector obtained after expanding the training multimodal electronic product recycling value representation feature map using the following optimization formula to obtain the optimized training multimodal electronic product recycling value representation feature vector; wherein, the optimization formula is:

[0048]

[0049] Where V is the feature vector representing the recycling value of the trained multimodal electronic products, v i It is the feature value at the i-th position of the feature vector representing the recycling value of the trained multimodal electronic products. It is the square of the 1-norm of the feature vector V representing the recycling value of the trained multimodal electronic products, ‖V‖2 -1 / 2 is the reciprocal of the square root of the 2-norm of the trained multimodal electronic product recycling value representation feature vector V, L is the length of the trained multimodal electronic product recycling value representation feature vector V, ε is the scaling hyperparameter, log represents the logarithmic function value to the base 2, and v′ i It is the feature value at the i-th position of the optimized trained multimodal electronic product recycling value representation feature vector.

[0050] Here, the low-rank structure of the norm of the training multimodal electronic product recycling value representation feature vector V is represented as a voting cluster of the feature value set aggregation of the training multimodal electronic product recycling value representation feature vector V. Each feature value of the training multimodal electronic product recycling value representation feature vector V undergoes canonical voting relative to the information perception framework. By aggregating the direction and scale of the feature distribution regression representation, feature values ​​belonging to the same regression class are mapped to similar locally canonical coordinate sets, thereby improving the aggregation effect of the feature set of the training multimodal electronic product recycling value representation feature vector V and improving the classification regression effect of the training multimodal electronic product recycling value representation feature vector V through the classifier, i.e., improving the training efficiency and accuracy of the classifier. In this way, the recycling value level of waste electronic products can be comprehensively assessed and judged based on the surface state visual semantic features and basic information key semantic features of the waste electronic products. This approach improves the efficiency and accuracy of waste electronic product recycling, thereby achieving effective resource utilization and sustainable environmental development.

[0051] In summary, an IoT system 100 based on intelligent recycling traceability based on embodiments of this application is explained, which can realize the recording and traceability of product-related data.

[0052] As described above, the IoT system 100 based on the intelligent recycling traceability of this application embodiment can be implemented in various terminal devices, such as servers with IoT algorithms based on the intelligent recycling traceability of this application embodiment. In one example, the IoT system 100 based on the intelligent recycling traceability of this application embodiment can be integrated into the terminal device as a software module and / or hardware module. For example, the IoT system 100 based on the intelligent recycling traceability of this application embodiment can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the IoT system 100 based on the intelligent recycling traceability of this application embodiment can also be one of many hardware modules of the terminal device.

[0053] Alternatively, in another example, the IoT system 100 based on the intelligent recycling traceability based on the embodiments of this application and the terminal device may also be separate devices, and the IoT system 100 based on the intelligent recycling traceability can be connected to the terminal device via wired and / or wireless networks, and transmit interactive information in accordance with the agreed data format.

[0054] Figure 3 This is a flowchart of an IoT method based on intelligent recycling traceability according to an embodiment of this application. Figure 3 As shown, the IoT method based on intelligent recycling traceability according to an embodiment of this application includes: S110, acquiring product-related data provided by smart tags installed on waste electronic products, wherein the product-related data includes basic information of the electronic products and images of the electronic products; S120, extracting features from the electronic product images using an electronic product surface state feature extractor based on a deep neural network model to obtain an electronic product surface state feature map; S130, performing adaptive attention enhancement processing on the electronic product surface state feature map to obtain an adaptive attention enhanced electronic product surface state feature map; S140, semantically encoding the basic information of the electronic products to obtain a semantic encoding feature vector of the basic information of the electronic products; S150, performing cross-modal feature fusion on the adaptive attention enhanced electronic product surface state feature map and the semantic encoding feature vector of the basic information of the electronic products to obtain multimodal electronic product recycling value characterization features; and S160, determining a recycling value level label based on the multimodal electronic product recycling value characterization features.

[0055] Here, those skilled in the art will understand that the specific operations of each step in the above-described IoT method based on intelligent recycling traceability have been referenced above. Figures 1 to 2 The IoT system 100 based on intelligent recycling and traceability is described in detail therein, and therefore, its repeated description will be omitted.

[0056] Figure 4 This is an application scenario diagram of an IoT system based on intelligent recycling and traceability, according to an embodiment of this application. For example... Figure 4 As shown, in this application scenario, firstly, product-related data (e.g., data from smart tags installed on the waste electronic products) is acquired. Figure 4 As shown in D), the product-related data includes basic information about the electronic product and an image of the electronic product. Then, the product-related data is input to a server deployed with an IoT algorithm based on intelligent recycling traceability (e.g., ...). Figure 4In the S shown, the server is able to process the product-related data using the IoT algorithm based on smart recycling traceability to obtain a classification result for a grade label representing the recycling value.

[0057] According to another aspect of this application, a non-volatile computer-readable storage medium is also provided, on which computer-readable instructions are stored, which, when executed by a computer, can perform the methods described above.

[0058] The program portion of a technology can be considered a "product" or "artifact" existing in the form of executable code and / or related data, and is involved in or implemented through a computer-readable medium. Tangible, permanent storage media can include memory or storage used by any computer, processor, or similar device or related module. For example, various semiconductor memories, tape drives, disk drives, or any similar device capable of providing storage functionality for software.

[0059] This application uses specific terms to describe embodiments of the application. Terms such as "first / second embodiment," "an embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of the application. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Furthermore, certain features, structures, or characteristics in one or more embodiments of the application can be appropriately combined.

[0060] Furthermore, those skilled in the art will understand that aspects of this application can be described and illustrated through several patentable types or situations, including any new and useful combination of processes, machines, products, or substances, or any new and useful improvements thereof. Accordingly, aspects of this application can be implemented entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. All of the above hardware or software may be referred to as a “data block,” “module,” “engine,” “unit,” “component,” or “system.” Furthermore, aspects of this application may manifest as a computer product located on one or more computer-readable media, the product including computer-readable program code.

[0061] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in a common dictionary shall be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not as having an idealized or highly formalized meaning, unless expressly defined herein.

[0062] The foregoing description is a illustrative description of the present application and should not be construed as limiting it. Although several exemplary embodiments of the present application have been described, those skilled in the art will readily understand that many modifications can be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all such modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the foregoing description is a illustrative description of the present application and should not be construed as limiting it to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. An IoT system based on intelligent recycling and traceability, characterized in that, include: A smart tag installation module is used to install smart tags on waste electronic products, wherein the smart tags are used to record product-related data of the waste electronic products; The cloud platform processing module is used to store and process the product-related data of the waste electronic products recorded by the smart tags, so as to perform recycling value analysis on the waste electronic products; The recycling station module is used to provide recycling services for the waste electronic products, and to obtain product-related data of the waste electronic products by scanning the smart tags, so as to classify, pack and transport the waste electronic products. as well as The waste electronic product processing module is used to dismantle, separate, and reuse the recycled waste electronic products, and update the product-related data of the waste electronic products by scanning the smart tags, and then feed the updated product-related data of the waste electronic products back to the cloud platform processing module.

2. The IoT system based on intelligent recycling and traceability according to claim 1, characterized in that, The cloud platform processing module includes: The product-related data acquisition unit is used to acquire product-related data provided by smart tags installed on the waste electronic products, wherein the product-related data includes basic information of the electronic products and images of the electronic products; An electronic product surface state feature extraction unit is used to extract features from the electronic product image using an electronic product surface state feature extractor based on a deep neural network model to obtain an electronic product surface state feature map. An electronic product surface state feature enhancement unit is used to perform adaptive attention enhancement processing on the electronic product surface state feature map to obtain an adaptive attention enhanced electronic product surface state feature map. The basic information semantic encoding unit for electronic products is used to semantically encode the basic information of the electronic products to obtain the semantic encoding feature vector of the basic information of electronic products. A cross-modal feature fusion unit is used to perform cross-modal feature fusion on the adaptive attention-enhanced electronic product surface state feature map and the electronic product basic information semantic encoding feature vector to obtain multimodal electronic product recycling value representation features; and The recycling value detection unit is used to determine the recycling value grade label based on the recycling value characterization characteristics of the multimodal electronic products.

3. The IoT system based on intelligent recycling and traceability according to claim 2, characterized in that, The deep neural network model is a convolutional neural network model.

4. The IoT system based on intelligent recycling and traceability according to claim 3, characterized in that, The electronic product surface state feature enhancement unit is used to: pass the electronic product surface state feature map through an adaptive attention module to obtain the adaptive attention enhanced electronic product surface state feature map.

5. The IoT system based on intelligent recycling and traceability according to claim 4, characterized in that, The electronic product surface state feature enhancement unit is configured to: process the electronic product surface state feature map through the adaptive attention module using the following adaptive enhancement formula to obtain the adaptive attention-enhanced electronic product surface state feature map; wherein, the adaptive enhancement formula is: F′=A′⊙F A=σ(W a *v+B a ) v = pool(F) Where F is the surface state feature map of the electronic product, pool(·) represents global mean pooling of each feature matrix along the channel dimension in the feature map, v is the channel feature vector of the surface state feature map of the electronic product, and W a and B a Here, σ represents the weights and biases of the convolutional layer, σ is the activation function, and A is the convolutional feature vector of the channel feature vector. i A is the feature value at each position in the convolutional feature vector. ′ It is a weight vector, ⊙ is the positional dot product, F ′ It is the surface state feature map of the adaptive attention-enhanced electronic product.

6. The IoT system based on intelligent recycling and traceability according to claim 5, characterized in that, The cross-modal feature fusion unit is used to: process the adaptive attention-enhanced electronic product surface state feature map and the electronic product basic information semantic encoding feature vector using a meta-network-based cross-modal feature fusion unit to obtain a multimodal electronic product recycling value representation feature map as the multimodal electronic product recycling value representation feature.

7. The IoT system based on intelligent recycling and traceability according to claim 6, characterized in that, The cross-modal feature fusion unit includes: The first convolutional subunit is used to pass the semantically encoded feature vector of the basic information of the electronic product through a point convolutional layer to obtain the first convolutional feature vector. The first correction subunit is used to pass the first convolutional feature vector through a correction linear unit based on the ReLU function to obtain a first corrected convolutional feature vector; The second convolutional subunit is used to pass the first modified convolutional feature vector through a point convolutional layer to obtain the second convolutional feature vector; The second correction subunit is used to pass the second convolutional feature vector through a sigmoid-based correction linear unit to obtain a second corrected convolutional feature vector; and A fusion subunit is used to fuse the second modified convolutional feature vector with the adaptive attention-enhanced electronic product surface state feature map to obtain the multimodal electronic product recycling value representation feature map.

8. The IoT system based on intelligent recycling and traceability according to claim 7, characterized in that, The recycling value detection unit is used to: pass the recycling value characterization feature map of the multimodal electronic product through a classifier to obtain a classification result, and the classification result is used to represent the recycling value level label.

9. The IoT system based on intelligent recycling and traceability according to claim 8, characterized in that, It also includes a training module for training the electronic product surface state feature extractor based on the convolutional neural network model, the adaptive attention module, the cross-modal feature fusion unit based on the meta-network, and the classifier; The training module includes: A training data acquisition unit is used to acquire training data, which includes training product-related data provided by smart tags installed on the waste electronic products, wherein the training product-related data includes basic information of the training electronic products and images of the training electronic products. The training electronic product surface state feature extraction unit is used to extract features from the training electronic product image through the electronic product surface state feature extractor based on the convolutional neural network model to obtain the training electronic product surface state feature map; A training electronic product surface state feature enhancement unit is used to pass the training electronic product surface state feature map through the adaptive attention module to obtain a training adaptive attention enhanced electronic product surface state feature map. A training electronic product basic information semantic encoding unit is used to perform semantic encoding on the training electronic product basic information to obtain a training electronic product basic information semantic encoding feature vector. A cross-modal feature fusion unit is trained to process the trained adaptive attention-enhanced electronic product surface state feature map and the trained electronic product basic information semantic encoding feature vector using the meta-network-based cross-modal feature fusion unit to obtain a trained multimodal electronic product recycling value representation feature map. An optimization training unit is used to perform feature aggregation optimization on the training multimodal electronic product recycling value representation feature vector obtained after expanding the training multimodal electronic product recycling value representation feature map to obtain an optimized training multimodal electronic product recycling value representation feature vector. A classification loss calculation unit is used to pass the optimized trained multimodal electronic product recycling value representation feature vector through the classifier to obtain a classification loss function value; and The loss training unit is used to train the electronic product surface state feature extractor based on the convolutional neural network model, the adaptive attention module, the cross-modal feature fusion unit based on the meta-network, and the classifier using the classification loss function value.

Images