@article{SU2026, 
author = {Fang SU and Jialiang ZHANG and Meirong LEI and Chensheng WANG},
title = {Digital twin experimental system design for an intelligent dry coal separator},
year = {2026},
journal = {Experimental Technology and Management},
volume = {43},
number = {9},
pages = {122-130},
keywords = {intelligent dry coal separator, digital twin, NX MCD, machine vision, virtual debugging},
url = {https://www.sciopen.com/article/10.16791/j.cnki.sjg.2026.09.015},
doi = {10.16791/j.cnki.sjg.2026.09.015},
abstract = {ObjectiveIntelligent dry coal separators are essential teaching equipment for electromechanical engineering and mineral processing majors within the framework of emerging engineering education. However, prominent drawbacks existed in physical experimental teaching. The high procurement and maintenance costs prevented universities from equipping themselves with complete production-grade units; high-speed separation operations posed safety risks; repeated parameter adjustments accelerated equipment wear; diverse operating conditions with variable particle sizes and material shapes could not be fully reproduced using physical prototypes. Conventional teaching methods, relying on textbooks and video observation, failed to cultivate students’ capabilities in machine vision recognition, PLC logic debugging, and material trajectory analysis. To address these limitations, an NX MCD–based digital twin experimental system was developed in this paper to provide a low-cost, safe, and repeatable virtual training platform for intelligent coal separation courses.MethodsFirst, the overall experimental scheme was formulated using the modular layout and separation mechanism of an intelligent dry coal separator. The system was divided into feeding, visual detection, air-blowing separation, dust removal, and electrical control units, and its functional architecture and technical indicators were defined. Second, a high-fidelity digital twin model was constructed using NX MCD. Mechanical parts were assembled in a modular manner, and rigid bodies, collision bodies, and kinematic pairs were configured for conveyor belts and sliding columns, which replaced pneumatic injectors. A virtual vision module was deployed with a preset camera resolution of 1280 px×720 px, a frame rate of 30 Hz, a horizontal FOV of 60°, and a 1000 lx ring lighting. Built-in image processing functions, including grayscale conversion, Gaussian filtering, Sobel edge detection, and threshold segmentation, were adopted, and material feature thresholds of gray value, particle size, and circularity were established to distinguish coal and gangue automatically. Third, a PLC-centered control system was designed, and the OPC UA protocol was adopted to establish real-time bidirectional data exchange between the PLC and the twin model. A virtual commissioning module was developed to realize motor start–stop, visual signal reception, trigger latching, delay timing, and reciprocating motion control of the sliding column. Finally, controlled variable experiments were performed to investigate the effects of ore size, shape, sensor trigger time, and collision position, with projectile vertex height and separation distance serving as evaluation indicators.ResultsThe twin model achieves multidimensional mapping of the separator’s geometry, motion, and control rules, and the virtual vision system stably identifies coal and gangue under simulated dust and light interference. PLC control logic can be fully verified without physical hardware. Experimental results show that 1) among ore specimens measuring 50 mm×50 mm, 75 mm×75 mm, and 100 mm×100 mm, the 75 mm×75 mm specimen is best matched to the actuator parameters and achieves the highest projectile trajectory. 2) Square ore specimens exhibit more stable stress distributions and superior separation performance than circular and 24-sided specimens. 3) An advance trigger time of 1.0 s synchronizes sliding column collision with material arrival and produces a higher ejection height than advance trigger times of 1.2 s and 1.5 s. 4) Collision on the ore front converts more impact energy into vertical velocity and results in an optimal separation effect.ConclusionsThis NX MCD–based digital twin experimental system effectively eliminates the teaching bottlenecks associated with costly equipment, difficult physical commissioning, and irreproducible complex working conditions. It supports hierarchical training covering structural cognition, principle demonstration, and parameter optimization, enabling visualized, quantifiable, and repeatable virtual experiments. The experimental conclusions provide direct calibration references for PLC delay parameters and sliding column layouts. This research innovates the experimental teaching mode of intelligent separation equipment, offering theoretical support for developing digital twin training platforms for mining machinery and demonstrating substantial educational and engineering application potential.}
}