CN-121972417-A - Three-dimensional recoverable industrial waste's letter sorting system

CN121972417ACN 121972417 ACN121972417 ACN 121972417ACN-121972417-A

Abstract

The invention relates to a three-dimensional recyclable industrial waste sorting system and a sorting method thereof, wherein the three-dimensional recyclable industrial waste sorting system comprises an input device, an identification device, a sorting device, an output device and a control platform which are sequentially arranged; the sorting system adopts three-dimensional multi-floor layout, and industrial waste is transmitted among different floors through a vertical hoistway. According to the invention, through multi-floor three-dimensional design, industrial waste is put in, identified, sorted and compressed and output are distributed in different vertical spaces, so that the planar occupied area of the system is greatly reduced, and the overall material throughput is improved. The AI image recognition technology and the infrared positioning sensing technology are combined, so that the accurate and rapid recognition of the materials of industrial wastes is realized, and the accuracy of sorting actions is ensured. And the system has self-evolution capability and keeps high recognition rate for a long time through the feedback optimization module.

Inventors

NIE WEN
XIONG WEI
XIONG YUHAO
WU XINRONG
HU XUEQI
ZHANG LIHONG

Assignees

蔚复来(浙江)科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260108

Claims (6)

1. The three-dimensional recyclable industrial waste sorting system is characterized by comprising an input device, an identification device, a sorting device, an output device and a control platform which are sequentially arranged, wherein the sorting system adopts three-dimensional multi-floor layout, and industrial waste is transmitted among different floors through a vertical hoistway; The recognition device comprises an image acquisition module, a preprocessing module, a feature extraction module, a classification model module, a result output module and a feedback optimization module; the system comprises an image acquisition module, a preprocessing module, a characteristic extraction module, a classification model module, a feedback optimization module, a control platform, a classification model module and a feedback optimization module, wherein the image acquisition module is used for acquiring multi-view image data of industrial wastes in real time; The sorting device comprises a sorting conveyor belt, a collecting assembly, a position sensor, a swing arm assembly and a driving assembly, wherein the sorting conveyor belt is connected with a bottom platform of the vertical shaft and is used for conveying industrial wastes, the collecting assembly is arranged on one side of the sorting conveyor belt and is used for collecting the industrial wastes on the sorting conveyor belt, the position sensor is arranged on the sorting conveyor belt and is used for detecting the positions of the industrial wastes on the sorting conveyor belt, the swing arm assembly is arranged on the sorting conveyor belt and is used for guiding the industrial wastes on the sorting conveyor belt into the collecting assembly, and the driving assembly is used for driving the sorting conveyor belt and the swing arm assembly; The control platform is connected with the identification device and the sorting device and is used for controlling the swing arm assembly according to the sorting result output by the identification device, the position information detected by the sorting device and the speed information of the sorting conveyor belt, and guiding the industrial waste into the collecting assembly.
2. The three-dimensional, recyclable industrial waste sorting system of claim 1, wherein the input device comprises an input conveyor belt, and wherein the output conveyor belt is provided with a weighing module for weighing industrial waste on the input conveyor belt and sending weighing information thereof to the control platform.
3. The three-dimensional recyclable industrial waste sorting system according to claim 1, wherein the image acquisition module comprises a camera and an auxiliary light source, the image processing performed by the preprocessing module comprises graying processing, denoising processing and size standardization processing, and the key features extracted by the feature extraction module comprise shape features, material features and color features.
4. The three-dimensional recyclable industrial waste sorting system of claim 1, wherein the neural network model employed by the classification model module is a modified ResNet-50 model that draws attention to mechanisms; The classification result output by the classification model module comprises classification information and confidence level; If the confidence coefficient is more than or equal to 90%, the classification result is directly output, if the confidence coefficient is less than 90%, the secondary shooting is triggered, the shot picture is subjected to the processing again, and if the confidence coefficient is still less than 90%, the picture is marked as a state of 'to-be-manually rechecked', and is singly sorted.
5. The three-dimensional recyclable industrial waste sorting system according to claim 1, wherein the collection assembly comprises a rechecking box and a plurality of collection boxes, wherein the rechecking box is used for collecting industrial waste to be rechecked manually, the plurality of collection boxes are used for collecting different types of industrial waste, the openings of the plurality of collection boxes are provided with infrared sensors, and the infrared sensors are used for detecting whether the industrial waste enters the interior of the collection boxes and sending detection information to the control platform; the swing arm assemblies are arranged on one side of the sorting conveyor belt, far away from the collecting assemblies, and comprise swing arms with the same number as that of the boxes of the collecting assemblies, and the swing arms are rotatably arranged on a fixing frame of the sorting conveyor belt; the driving assembly comprises a servo motor, wherein the servo motor is connected with the control platform and is used for driving the swing arm according to a control signal sent by the control platform.
6. A method for sorting three-dimensional recyclable industrial waste, characterized in that the industrial waste is sorted based on the sorting system according to one of claims 1-5, The sorting method comprises the following steps: receiving and weighing industrial waste through an input device; Carrying out image acquisition, pretreatment, feature extraction and model classification on the industrial waste through the identification device, and outputting classification results; The industrial waste enters the sorting device through the vertical shaft, the position information of the industrial waste is detected through the position sensor, and the control platform calculates the swing arm time according to the sorting result and the position information and by combining the speed of the output conveyor belt; the control platform drives the swing arm according to the calculation result, and guides the industrial waste into a designated collection box; After the weight of the industrial waste in the collecting box reaches the set weight, the industrial waste is conveyed to the compression equipment through the output conveyor belt to be compressed.

Description

Three-dimensional recoverable industrial waste's letter sorting system Technical Field The invention relates to the technical field of AI waste sorting, in particular to a three-dimensional recyclable industrial waste sorting system. Background With the rapid development of the age, the number of recyclable industrial waste products has increased dramatically, causing serious environmental stress. At present, most of sorting treatment of recyclable industrial waste depends on manual work, the efficiency is low, and the sorting effect is not ideal. To address this problem, some sorting devices have been designed by some garbage disposal manufacturers to sort industrial waste, but these sorting devices all have certain drawbacks, in particular: 1) The method improves a certain sorting efficiency by the system operation, but still requires a large amount of fixed manpower to cooperate to finish the sorting work, so that the sorting cost is too high, and on the other hand, during the manual sorting process, the industrial waste is possibly classified by mistake due to various reasons (such as fatigue, inattention and the like), so that the efficiency and quality of resource recovery are influenced. 2) The method has high efficiency, but cannot specifically identify the materials of the industrial waste, is easy to cause waste and pollution, and has limited expansibility. Therefore, the development of the efficient and intelligent recyclable industrial waste sorting system has important significance in improving the sorting efficiency of industrial waste, reducing the resource waste and protecting the environment. Disclosure of Invention The invention aims to provide a three-dimensional recyclable industrial waste sorting system which solves the defects in the prior art. The invention solves the technical problems by adopting the following technical scheme: the three-dimensional recyclable industrial waste sorting system is characterized by comprising an input device, an identification device, a sorting device, an output device and a control platform which are sequentially arranged, wherein the sorting system adopts three-dimensional multi-floor layout, and industrial waste is transmitted among different floors through a vertical hoistway; The recognition device comprises an image acquisition module, a preprocessing module, a feature extraction module, a classification model module, a result output module and a feedback optimization module; the system comprises an image acquisition module, a preprocessing module, a characteristic extraction module, a classification model module, a feedback optimization module, a control platform, a classification model module and a feedback optimization module, wherein the image acquisition module is used for acquiring multi-view image data of industrial wastes in real time; The sorting device comprises a sorting conveyor belt, a collecting assembly, a position sensor, a swing arm assembly and a driving assembly, wherein the sorting conveyor belt is connected with a bottom platform of the vertical shaft and is used for conveying industrial wastes, the collecting assembly is arranged on one side of the sorting conveyor belt and is used for collecting the industrial wastes on the sorting conveyor belt, the position sensor is arranged on the sorting conveyor belt and is used for detecting the positions of the industrial wastes on the sorting conveyor belt, the swing arm assembly is arranged on the sorting conveyor belt and is used for guiding the industrial wastes on the sorting conveyor belt into the collecting assembly, and the driving assembly is used for driving the sorting conveyor belt and the swing arm assembly; The control platform is connected with the identification device and the sorting device and is used for controlling the swing arm assembly according to the sorting result output by the identification device, the position information detected by the sorting device and the speed information of the sorting conveyor belt, and guiding the industrial waste into the collecting assembly. Further, the input device comprises an input conveyor belt, a weighing module is arranged on the output conveyor belt and used for weighing industrial waste on the input conveyor belt and sending weighing information of the industrial waste to the control platform. The image acquisition module comprises a camera and an auxiliary light source, the image processing executed by the preprocessing module comprises gray processing, denoising processing and size standardization processing, and the key features extracted by the feature extraction module comprise shape features, material features and color features. Further, the neural network model adopted by the classification model module is an improved ResNet-50 model for introducing an attention mechanism; The classification result output by the classification model module comprises classification information and confidence level; If the confide