Intelligent solution for comprehensive utilization of construction waste
This project aims to address two core issues in the field of construction waste treatment: efficient crushing and resource utilization. By combining the discrete element method (DEM) and object detection algorithm for optimization and improvement, the project has designed an innovative construction waste crushing device, achieving efficient crushing and intelligent precise sorting of construction waste.
Competencies And Staff Development
National projects for training in the fields of artificial intelligence, critical technologies or unmanned aircraft systems
3year
Henan, China
(1)Crushing efficiency: The discrete element method (DEM) is used to simulate the mechanical model and crushing mechanism of the crushing device from a microscopic perspective, assisting the crushing work. Compared with conventional construction waste crushing devices, under ideal conditions, such as uniform material feeding and normal operation of various components inside the crusher, the maximum processing capacity can reach 300 tons/hour. Considering factors such as material unevenness and variations in impurity content that may be encountered in actual operations, the standard processing capacity is 250 tons per hour. When encountering harsh working environments or material characteristics (such as high humidity, containing a large amount of steel bars, etc.), the minimum processing capacity can reach 180 tons/hour. The crushing efficiency of conventional construction waste crushing devices has been improved by 6.8%, 9.2%, and 11.6%, respectively. (2)Sorting accuracy: BoVW feature statistics based on local constraints, using MobileNet The feature extraction structure of ShuffleNet and EfficientNet replaces Backbone, and the sorting purity of organic compounds such as concrete blocks, bricks, metals, and plastics is higher than 95%, 93%, 98%, and 90%, respectively. The missorting rate is lower than 3%, 4%, 1%, and 5%. Considering the complexity of plastic organic sorting, the missorting rate is slightly higher, but still within an acceptable range. (3)Detection efficiency: The improved object detection algorithm can achieve a detection speed of 50 frames per second, compared to SSD The detection speed of RFBNet and RetinaNet models was 117.4%, 100%, and 85.2% faster, respectively.
1.The project won the third prize of "Internet plus" Undergraduate Innovation and Entrepreneurship Competition in Henan Province, China 2. Project: "Research on the mechanical properties of prefabricated road base based on recycled aggregates from construction waste" has been completed
1. Patent 1: A construction waste crushing device for construction projects, patent number: ZL2020215200209 2. Patent 2: A cargo sorting device for logistics management, patent number: ZL2021217372647 3. Published a representative paper on the utilization of construction waste in prefabricated roads in the Chinese scientific journal database "Natural Science"
