Development And Application Of An Automatic Sorting System For Scrap Metal Materials Based On Self Consistency Of Dynamic Coupling

This paper establishes the self-consistent basic configuration of the system featuring information perception,information cognition and information decision for materials in accordance with its dynamic coupling correlation between random form,complex composition and real-time online,focuses on researching an automatic sorting system and its application methods and conducts systematical test to meet the needs of different tasks.This helps solve the problem of degraded utilization caused by rough sorting in the area of recycling and full dependence on manual work and,in particular,meet the demand for refined and efficient online analysis of high-quality scrap metal materials.First,a research on spectral information perception for complex shaped materials in online analysis is carried out and an automatic sorting system is developed.Materials are tracked and positioned through an identification module based on dualsensor fusion technology,after which laser-induced breakdown spectroscopy module is designed to efficiently perceive spectrum signal of dynamic materials.A composite material separation module is developed to achieve adaptive separation of material form.Through the coupling design of high-precision hardware and parallel software control system,the automated collaborative control of multiple modules in the system is completed,thereby achieving online spectral information perception of dynamic materials.Second,based on the coupling of multiple parameters,a systematic spectral response characterization method is established for spectral information mining.Select standard sample of two materials to analyze spectral response,and establish an optimization model by having systemic spectral response correlation analysis between net analyte signal and signal-to-background ratio at the different delay times.By studying what influences spectral response performance,the math model for systemic spectral response efficiency is established,achieving the evaluation of spectral response capability under a given coupling disturbance.Meanwhile,the boundary value of coupling disturbance can also be obtained on the basis of spectral response capability,providing a quantitative analysis method for the coupling spectral information.Third,research on classification analysis method of feature-based algorithm is conducted to establish information decision analysis method.Through studies on three data dimension reduction methods,namely spectral line calibration,principal component analysis and genetic algorithm,as well as comparison of classification modeling,genetic algorithm is found to have obvious advantages in selecting spectral lines in coupled spectral information.However,the verification results about the stability of the method show that while characteristic spectral lines are effectively selected,interference features are also randomly introduced,leading to differences in model prediction ability.The paper presents an adaptive characteristic spectral line screening algorithm based on the principle of genetic algorithm and physical properties of dynamic spectra.By analyzing classification model validation results,the adaptive feature screening algorithm effectively reduces the dimension of features and improves the ability to suppress disturbance.The predictive and generalization ability of the model for refined classification are enhanced.Finally,to meet the actual demand and fulfill the primary goal of refined material classification,test on sorting system and application method is conducted regarding different materials.Three typical high-value scrap metal materials,namely stainless steel,aluminum alloy and titanium alloy,are chosen in the form of flakes and scattered scrap metal to establish analysis methods in accordance with different targets and complete verification test through different modes of systemic online dynamic test.The accuracy of online analysis for refined classification of materials is good.The related work provides a reference for engineering application of automated and refined online sorting of the scrap metal materials.