Using Artificial Intelligence and 3D Imaging to Provide Real-time Particle Size Analysis on Haul Trucks

Continuing blast optimization is the cornerstone of any mine-to-mill approach, and particle size analysis is a key metric for meeting blast targets to reduce overall energy consumption and operating costs. To ensure the statistical validity of particle size data, it is important to sample material frequently. In most mines and quarries, the blasted material is loaded onto haul trucks and then transferred to the crusher; measuring the particle size of each haul truck load therefore provides an adequate sampling rate. We propose a particle size analysis solution for haul trucks that does not impact current operational workflows. Using artificial intelligence (AI) and 3D imaging technology, our solution reliably localizes and identifies haul trucks as they drive by, then accurately calculates the particle size distribution of the visible material in each truck load. In this paper, we identify the challenges of designing and implementing the proposed solution and explain the strategies we employ to overcome these. We also discuss a method for boulder detection that leverages the existing deep learning architecture to significantly reduce crusher downtime caused by oversized rocks. Finally, we present field data from a trial installation. To evaluate each change made to the blast design, it is necessary to measure the effect on particle size. An accurate particle size analysis solution is therefore the first step towards optimizing blasting practices and reducing energy consumption throughout the comminution circuit.