Intelligent sorting system and method for waste textiles

Abstract

The invention discloses an intelligent sorting system and method for waste textiles. The intelligent sorting system comprises a multi-modal information acquisition module, a self-adaptive learning module, a material recognition optimization module and a sorting execution module. The data acquisition module is used for acquiring multi-dimensional data of the waste textiles, and the multi-dimensional data comprises thermal infrared image data, visible light image data and material surface texture feature data; the adaptive learning module is used for constructing a dynamic adjustment model according to the multi-dimensional data in the multi-modal information acquisition module; the invention relates to the technical field of waste textile recycling. According to the intelligent sorting system and method for the waste textiles, real-time fusion of thermal infrared image data, visible light image data and material surface texture feature data is achieved by introducing the multi-modal information acquisition module, the problem that single-modal information recognition precision is insufficient is solved, a dynamic adjustment model is constructed through the self-adaptive learning module, and the recognition accuracy is improved. And the response speed and accuracy of the system to diversified waste textiles are improved.

Description

Technical Field

[0001] This invention relates to the field of waste textile recycling technology, specifically to an intelligent sorting system and method for waste textiles. Background Technology

[0002] With the continuous development of my country's economy and society, the amount of waste textiles generated continues to increase. Currently, with the rise of environmental awareness, research on the recycling and utilization of waste textiles has gradually begun. From a practical perspective, waste textiles in daily life are diverse in type and material, and are highly mixed. There is still significant room for improvement in the sorting efficiency and accuracy of waste textiles.

[0003] Reference patent publication number "CN118719623B" discloses an intelligent identification system and method for waste textiles based on image processing, including: a thermal infrared identification module, used to perform thermal infrared image identification on waste textiles in sorting operation to obtain the presence state information of non-fibrous materials inside the waste textiles; and a sorting operation adjustment module, used to adjust the operation state of the sorting operation based on the presence state information of non-fibrous materials.

[0004] As shown in the above technology, the existing technology mainly relies on single-modal images, thermal infrared or visible light in specific operation stages, and fails to fully realize the real-time fusion and collaborative analysis of multimodal information. This may lead to limitations in the recognition accuracy of complex materials during the sorting process. In addition, the adjustment of the sorting operation status is mainly based on preset rules, and the adaptive learning ability is lacking. When faced with diverse waste textiles, there may be response delays or misjudgments. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent sorting system and method for waste textiles, which solves the problems that existing intelligent sorting systems and methods for waste textiles still have certain deficiencies in multimodal information fusion, adaptive learning capabilities, and high-precision sorting for complex materials.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a waste textile intelligent sorting system includes a multimodal information acquisition module, an adaptive learning module, a material identification and optimization module, and a sorting execution module; The multimodal information acquisition module is used to acquire multi-dimensional data of waste textiles, including thermal infrared image data, visible light image data, and material surface texture feature data; the adaptive learning module is used to construct a dynamically adjusted model based on the multi-dimensional data in the multimodal information acquisition module; the material identification optimization module is used to classify and analyze the complex materials of waste textiles in conjunction with the dynamically adjusted model and generate material classification labels; the sorting execution module is used to receive the material classification labels and drive the sorting equipment to complete the actual sorting operation of waste textiles. The output of the multimodal information acquisition module is connected to the input of the adaptive learning module; the output of the adaptive learning module is connected to the input of the material recognition optimization module; and the output of the material recognition optimization module is connected to the input of the sorting execution module.

[0007] Preferably, the multimodal information acquisition module includes a thermal imaging unit, an optical imaging unit, and a texture detection unit; The thermal imaging unit is used to capture thermal infrared image data of waste textiles, which includes the surface temperature distribution and thermal conductivity characteristics of the textiles; the optical imaging unit is used to capture visible light image data of waste textiles, which includes the color, shape and edge characteristics of the textiles; and the texture detection unit is used to acquire surface texture feature data of waste textiles, which includes fiber arrangement direction, roughness and gloss. The outputs of the thermal imaging unit, optical imaging unit, and texture detection unit are all connected to the input of the adaptive learning module.

[0008] Preferably, the adaptive learning module includes a data fusion processing unit and a dynamic adjustment unit; The data fusion processing unit is used to integrate thermal infrared image data, visible light image data, and surface texture feature data, and generate a comprehensive feature vector through a weighted algorithm; the dynamic adjustment unit is used to train a deep learning network based on the comprehensive feature vector, and continuously update the network parameters according to real-time input data, and output a dynamically adjusted model. The output of the data fusion processing unit is connected to the input of the dynamic adjustment unit.

[0009] Preferably, the material identification optimization module includes a classification label generation unit and a material property analysis unit; The classification label generation unit is used to call the dynamic adjustment model to classify the materials of waste textiles and generate material classification labels; the material property analysis unit is used to further analyze the physical and chemical properties of the materials based on the material classification labels and generate a material property report. The output of the classification label generation unit is connected to the input of the material property analysis unit.

[0010] Preferably, the sorting execution module includes a sorting instruction generation unit and a mechanical action execution unit; The sorting instruction generation unit is used to receive material classification labels and generate corresponding sorting instructions; the mechanical action execution unit is used to drive the sorting equipment to complete the sorting operation of waste textiles according to the sorting instructions. The output of the sorting instruction generation unit is connected to the input of the mechanical action execution unit.

[0011] Preferably, the classification label generation unit further includes: The probability values ​​of material classification for waste textiles are obtained based on the dynamic adjustment model; a classification probability threshold P0 is set; if the probability value of a certain material classification is greater than or equal to P0, it is marked as the final classification result; the final classification result is encoded to generate a material classification label.

[0012] This invention also discloses an intelligent sorting method for waste textiles, which specifically includes the following steps: Step 1: The intelligent sorting system for waste textiles first completes multi-dimensional data collection of the target items through the multi-modal information acquisition module; Step 2: After preliminary processing, the data collected in Step 1 is transmitted to the data fusion processing unit, which integrates data from different modalities into a comprehensive feature vector using a weighted algorithm. Step 3: Dynamically adjust the unit to train the deep learning network based on the comprehensive feature vector; Step 4: The classification label generation unit in the material recognition optimization module calls the dynamically adjusted model to classify waste textiles by material. Step 5: After receiving the material classification label, the sorting instruction generation unit in the sorting execution module generates sorting instructions based on the location information of the target sorting area.

[0013] Preferably, in step four, the material identification optimization module generates and analyzes the characteristics of material classification labels. For blended materials, it can not only identify the main components but also analyze the proportion of minor components, thereby generating a more accurate material characteristic report.

[0014] Beneficial effects This invention provides an intelligent sorting system and method for waste textiles. Compared with the prior art, it has the following advantages: 1. The intelligent sorting system and method for waste textiles achieves real-time fusion of thermal infrared image data, visible light image data and material surface texture feature data by introducing a multimodal information acquisition module, which solves the problem of insufficient recognition accuracy of single-modal information. By constructing a dynamic adjustment model through an adaptive learning module, the system's response speed and accuracy to diverse waste textiles are improved.

[0015] 2. The intelligent sorting system and method for waste textiles generates material classification labels and analyzes material characteristics through a material identification optimization module. The classification label generation unit calls a dynamic adjustment model to classify the materials of waste textiles. The material characteristic analysis unit further analyzes the physical and chemical properties of the materials based on the material classification labels and generates a material characteristic report, thereby improving the identification capability of complex materials.

[0016] 3. The intelligent sorting system and method for waste textiles completes efficient sorting operations by combining sorting instructions and mechanical actions through a sorting execution module. The sorting instruction generation unit receives material classification labels and generates sorting instructions, which include the location information of the target sorting area and sorting priority rules. The mechanical action execution unit plans the motion path of the robotic arm according to the sorting instructions and drives the robotic arm to move along the planned path to the target sorting area to complete the sorting operation, which greatly improves sorting efficiency. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the module structure of the system of the present invention; Figure 2 This is a flowchart illustrating the workflow of the multimodal information acquisition module of the present invention. Figure 3 This is a flowchart of the intelligent sorting method for waste textiles of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figures 1-3 This intelligent sorting system for waste textiles offers two technical solutions: The first implementation method includes a multimodal information acquisition module, an adaptive learning module, a material recognition optimization module, and a sorting execution module. The multimodal information acquisition module is used to acquire multidimensional data of waste textiles, including thermal infrared image data, visible light image data, and material surface texture feature data; the adaptive learning module is used to build a dynamically adjusted model based on the multidimensional data in the multimodal information acquisition module; the material identification and optimization module is used to classify and analyze the complex materials of waste textiles and generate material classification labels by combining the dynamic adjustment model; the sorting execution module is used to receive the material classification labels and drive the sorting equipment to complete the actual sorting operation of waste textiles. The output of the multimodal information acquisition module is connected to the input of the adaptive learning module; the output of the adaptive learning module is connected to the input of the material recognition optimization module; and the output of the material recognition optimization module is connected to the input of the sorting execution module.

[0020] The multimodal information acquisition module includes a thermal imaging unit, an optical imaging unit, and a texture detection unit. The thermal imaging unit is used to capture thermal infrared image data of waste textiles, which includes the surface temperature distribution and thermal conductivity characteristics of the textiles. The optical imaging unit is used to capture visible light image data of waste textiles, which includes the color, shape, and edge features of the textiles. The texture detection unit is used to acquire surface texture feature data of waste textiles, which includes fiber orientation, roughness, and gloss. The outputs of the thermal imaging unit, the optical imaging unit, and the texture detection unit are all connected to the input of the adaptive learning module.

[0021] The adaptive learning module includes a data fusion processing unit and a dynamic adjustment unit. The data fusion processing unit integrates thermal infrared image data, visible light image data, and surface texture feature data, and generates a comprehensive feature vector through a weighted algorithm. The dynamic adjustment unit trains a deep learning network based on the comprehensive feature vector and continuously updates the network parameters according to real-time input data, outputting a dynamically adjusted model. The output of the data fusion processing unit is connected to the input of the dynamic adjustment unit.

[0022] The material identification optimization module includes a classification label generation unit and a material property analysis unit. The classification label generation unit is used to call the dynamic adjustment model to classify the materials of waste textiles and generate material classification labels. The material property analysis unit is used to further analyze the physical and chemical properties of the materials based on the material classification labels and generate a material property report. The output of the classification label generation unit is connected to the input of the material property analysis unit.

[0023] The second implementation differs from the first implementation in that the sorting execution module includes a sorting instruction generation unit and a mechanical motion execution unit. The sorting instruction generation unit receives material classification labels and generates corresponding sorting instructions. The mechanical motion execution unit drives the sorting equipment to complete the sorting operation of waste textiles according to the sorting instructions. The output end of the sorting instruction generation unit is connected to the input end of the mechanical motion execution unit.

[0024] The classification label generation unit also includes: obtaining the material classification probability value of waste textiles according to the dynamic adjustment model; setting the classification probability threshold P0; if the probability value of a certain material classification is greater than or equal to P0, then marking it as the final classification result; encoding the final classification result to generate a material classification label.

[0025] This invention also discloses an intelligent sorting method for waste textiles, which specifically includes the following steps: Step 1: The intelligent sorting system for waste textiles first completes multi-dimensional data collection of the target items through a multi-modal information acquisition module. The thermal imaging unit captures thermal infrared image data of the waste textiles, including surface temperature distribution and thermal conductivity characteristics. The optical imaging unit acquires visible light image data of the waste textiles, covering color, shape, and edge features. The texture detection unit is responsible for collecting surface texture feature data such as fiber alignment direction, roughness, and gloss. Step 2: After preliminary processing, the above data is transmitted to the data fusion processing unit, which integrates the data from different modalities into a comprehensive feature vector through a weighted algorithm.

[0026] Step 3: The dynamic adjustment unit trains a deep learning network based on the comprehensive feature vector. During training, the initial error is defined using the cross-entropy loss function, and the network is iteratively trained. In each iteration, the error gradient of the input data samples is calculated, and a decision tree is fitted to determine the optimal correction value. This process is repeated until the preset maximum number of iterations is reached. The resulting deep learning network is the dynamic adjustment model, which can update parameters in real time to adapt to the diverse needs of waste textiles.

[0027] Step 4: The classification label generation unit in the material identification and optimization module calls the dynamic adjustment model to classify waste textiles by material. Based on the material classification probability value output by the dynamic adjustment model, a classification probability threshold P0 is set. If the probability value of a certain material classification is greater than or equal to P0, it is marked as the final classification result and a material classification label is generated. The material characteristic analysis unit further combines the database to retrieve the physical and chemical characteristics corresponding to the material type and generates a detailed material characteristic report.

[0028] Step 5: After receiving the material classification label, the sorting instruction generation unit in the sorting execution module generates sorting instructions based on the location information of the target sorting area. When generating instructions, the system will give priority to the material type and the distance to the sorting area. The mechanical motion execution unit plans the movement path of the robotic arm according to the sorting instructions, including the starting position, the ending position and the obstacle avoidance strategy. The robotic arm moves along the planned path to the target sorting area to complete the actual sorting operation of waste textiles.

[0029] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0030] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent sorting system for waste textiles, characterized in that: It includes a multimodal information acquisition module, an adaptive learning module, a material recognition optimization module, and a sorting execution module; The multimodal information acquisition module is used to acquire multi-dimensional data of waste textiles, including thermal infrared image data, visible light image data, and material surface texture feature data; the adaptive learning module is used to construct a dynamically adjusted model based on the multi-dimensional data in the multimodal information acquisition module; the material identification optimization module is used to classify and analyze the complex materials of waste textiles in conjunction with the dynamically adjusted model and generate material classification labels; the sorting execution module is used to receive the material classification labels and drive the sorting equipment to complete the actual sorting operation of waste textiles. The output of the multimodal information acquisition module is connected to the input of the adaptive learning module; the output of the adaptive learning module is connected to the input of the material recognition optimization module; and the output of the material recognition optimization module is connected to the input of the sorting execution module.

2. The intelligent sorting system for waste textiles according to claim 1, characterized in that: The multimodal information acquisition module includes a thermal imaging unit, an optical imaging unit, and a texture detection unit; The thermal imaging unit is used to capture thermal infrared image data of waste textiles, which includes the surface temperature distribution and thermal conductivity characteristics of the textiles; the optical imaging unit is used to capture visible light image data of waste textiles, which includes the color, shape and edge characteristics of the textiles; and the texture detection unit is used to acquire surface texture feature data of waste textiles, which includes fiber arrangement direction, roughness and gloss. The outputs of the thermal imaging unit, optical imaging unit, and texture detection unit are all connected to the input of the adaptive learning module.

3. The intelligent sorting system for waste textiles according to claim 1, characterized in that: The adaptive learning module includes a data fusion processing unit and a dynamic adjustment unit; The data fusion processing unit is used to integrate thermal infrared image data, visible light image data, and surface texture feature data, and generate a comprehensive feature vector through a weighted algorithm; the dynamic adjustment unit is used to train a deep learning network based on the comprehensive feature vector, and continuously update the network parameters according to real-time input data, and output a dynamically adjusted model. The output of the data fusion processing unit is connected to the input of the dynamic adjustment unit.

4. The intelligent sorting system for waste textiles according to claim 1, characterized in that: The material identification and optimization module includes a classification label generation unit and a material property analysis unit; The classification label generation unit is used to call the dynamic adjustment model to classify the materials of waste textiles and generate material classification labels; the material property analysis unit is used to further analyze the physical and chemical properties of the materials based on the material classification labels and generate a material property report. The output of the classification label generation unit is connected to the input of the material property analysis unit.

5. The intelligent sorting system for waste textiles according to claim 1, characterized in that: The sorting execution module includes a sorting instruction generation unit and a mechanical motion execution unit; The sorting instruction generation unit is used to receive material classification labels and generate corresponding sorting instructions; the mechanical action execution unit is used to drive the sorting equipment to complete the sorting operation of waste textiles according to the sorting instructions. The output of the sorting instruction generation unit is connected to the input of the mechanical action execution unit.

6. The intelligent sorting system for waste textiles according to claim 1, characterized in that: The classification label generation unit further includes: The probability values ​​of material classification for waste textiles are obtained based on the dynamic adjustment model; a classification probability threshold P0 is set; if the probability value of a certain material classification is greater than or equal to P0, it is marked as the final classification result; the final classification result is encoded to generate a material classification label.

7. A method for intelligent sorting of waste textiles, implemented by an intelligent sorting system for waste textiles as described in any one of claims 1-6, characterized in that: Specifically, the following steps are included: Step 1: The intelligent sorting system for waste textiles first completes multi-dimensional data collection of the target items through the multi-modal information acquisition module; Step 2: After preliminary processing, the data collected in Step 1 is transmitted to the data fusion processing unit, which integrates data from different modalities into a comprehensive feature vector using a weighted algorithm. Step 3: Dynamically adjust the unit to train the deep learning network based on the comprehensive feature vector; Step 4: The classification label generation unit in the material recognition optimization module calls the dynamically adjusted model to classify waste textiles by material. Step 5: After receiving the material classification label, the sorting instruction generation unit in the sorting execution module generates sorting instructions based on the location information of the target sorting area.

8. The intelligent sorting method for waste textiles according to claim 7, characterized in that: In step four, the material identification and optimization module generates and analyzes the characteristics of material classification labels. For blended materials, it can not only identify the main components but also analyze the proportion of minor components, thereby generating a more accurate material characteristic report.

Images