Research On Rapid Identification And Sorting System Of Community Household Waste

As the amount of urban household waste in China continues to increase,problems such as difficulty in garbage classification,low accuracy,and low efficiency in promotion have emerged.Traditional garbage classification equipment,due to its low level of intelligence and incomplete functions,poses a huge obstacle to garbage classification at the front-end sorting stage.The front-end classification is not clear,resulting in the back-end processing link needs to be sorted after the second classification.With the continuous development of artificial intelligence technology,its application scope in the protection of the ecological environment has also been expanding.This paper proposes a rapid identification,sorting and sorting system for community household garbage,which gradually identifies and classifies community household garbage in the front and back end,and then transfers to different posttreatment processes,so as to reduce the cost and pressure of disposing mixed waste.The system mainly includes the following three key research contents.A community garbage classification system based on the volume and bulk density was studied.The rapid classification technology of bulk density method is mainly used in the front end to screen out the wet garbage that cannot be recognized and sorted,so as to achieve the recognition goal of "no dry in the wet".Suitable volumetric weight threshold division rules are explored as the theoretical basis,and the sorting logic applicable to this system is designed.The overall design scheme of the system is proposed,the volume detection module is built,and an accuracy compensation plan is designed and verified through experiments.An experimental prototype is built and tested,and data is organized and analyzed for sample volume,bulk density detection results,and classification judgments.The test results show that the laser slicing volume and bulk density classification recognition technology system can achieve the functional requirements of this project and accurately identify multiple types of garbage.A machine vision recognition technology of community household garbage based on YOLOv5 model was studied.Machine vision recognition technology is mainly used in the back end to identify the recyclables in the dry waste after screening.A dataset for model training was created,and image information was labeled.The data set of model training was made and marked with image information,and the YOLOv5 s model with garbage type recognition function was trained based on the training set.The index results of model training are output and analyzed,and the training effect of the model is tested by using the test set picture material and the real shot picture material.The experimental results show that the target detection model fully learns the image features of various types of garbage in the training process,and the average accuracy of four-classification garbage recognition is high.The model trained in this paper has certain generalization and robustness,and meets the functional requirements of garbage sorting and recycling application scenarios.The sorting technology of community household garbage based on mechanical arm was studied.Sorting technology to undertake machine vision recognition technology,automatic sorting and classification of garbage after recognition.Based on the target detection model,the trained model is used to classify and locate the images collected by the camera,and the detection results are sent to the manipulator controller for sorting.Sorting,the manipulator kinematics analysis and simulation,camera calibration.The sorting experiment platform was built to verify the sorting system and test the sorting effect.The experimental results show that the target detection model has strong prediction ability after training.The detection success rate of the robot arm sorting system is 86.5%,and the sorting success rate is 85%,which meets the sorting function requirements in the garbage sorting and recycling application scenario.