Study On Complex Particle Swarm Space Image Reconstruction And Screening Performance Evaluation In Screening Process

In the ore beneficiation process,screening is a crucial step.If the screening performance cannot be monitored in a timely manner,it will directly affect the mineral separation.The core research direction to solve this problem is to develop a fast,high-precision,and highly stable intelligent screening particle size detection method.To address this issue,this thesis focuses on monitoring the screening performance of screening equipment,and builds a mineral particle size image recognition and analysis system,accurate measurement system,and visualization feedback system.It uses a linear laser profile sensor to reconstruct the three-dimensional spatial size of complex particle groups,discusses the particle size distribution characteristics under dynamic machine vision measurement,develops particle size composition analysis algorithms and screening performance evaluation algorithms,sets up a laboratory semi-industrial screening system for screening experiments,and conducts research on complex particle group spatial image reconstruction and screening performance evaluation during the screening process.We have built a mineral particle size image recognition and analysis system,clarified the mechanism of depth image processing,developed a particle capture algorithm for particle size composition of particle groups,and established a model to convert particle spatial size to actual size.We have clarified the measurement characteristics of 2D/3D machine vision and explored the measurement accuracy of linear laser profile sensors for irregular particles.The results show that using a linear laser profile sensor and depth image processing algorithm has a high advantage in obtaining particle size information,with low environmental impact,high measurement accuracy,and accurate particle size analysis for the entire particle group.We have analyzed and discussed the particle size composition measurement results under different representation methods,among which the equal-volume sphere particle size representation method performs the best,with identification accuracy greater than 90%for each particle size level,and up to 96%.Using a combined method of dynamic visual measurement and image analysis,we have clarified the impact of moisture on particle size distribution measurement.The results show that when the external water content of the feed is below 6%,moisture has little effect on the overall particle size distribution measurement.We have discussed the distribution characteristics of different unit area coal feed rates and found that appropriate feeding rates need to be controlled during particle size distribution measurement.The experimental results indicate that the best effect is achieved with a coal feed rate of 0.75 kg/m~2for a dense distribution.We have also investigated the impact of particle size range on particle size distribution measurement,and found that an increase in the measurement range will affect particle size distribution recognition.The maximum recognition particle size range for this method can be within 25-3mm or 50-6mm,and a reduction in the measurement interval scale will distribute the"mismatch error"to each interval without affecting the final measurement of particle size distribution.We have built a self-evaluation system for screening performance,developed particle size composition analysis and screening performance evaluation algorithms based on the Visual Studio platform,and achieved asynchronous acquisition and analysis of dynamic and static images.We conducted semi-industrial application research on screening experiments,and explored the impact mechanisms of processing volume,mineral particle size composition,and mineral species on screening performance evaluation indicators during the screening process.The results show that compared with traditional screening methods,the self-evaluation system for screening performance has a screening efficiency difference of less than 3%and a mismatched material content difference of less than 1%,achieving the expected evaluation effect.Based on this,we proposed an engineering application solution.