Scholarlink

This study aims to develop a support tool to beneficiation process for mineMine development projects by leveraging machine learningMachine learning on features of mineral particles obtained with electron microscopy and X-ray spectroscopy from a minimal quantity of ore samples. In this paper, a methodology to construct a predictive modelModel for flotationFlotation outcomes, including gradeGrade and recovery rateRecovery Rates under the same ore and flotationFlotation condition is proposed as a primary study. In the experiment, copper oreCopper ore feed subjected to various grindingGrinding conditions, along with the concentratesConcentrate and tailings obtained from flotationFlotation tests, were used. Training data for particle features were obtained using Mineral Liberation Analyzer and a predictive modelModel was subsequently constructed using machine learningMachine learning and evaluated for prediction accuracy. As a result, the predictions by the method were closer to the true values for test data under grindingGrinding conditions not included in the training dataset. This study contributes to solving the challenge of determining effective operating conditionsOperating conditions with a limited number of preliminary tests.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Primary Study on Predicting Flotation Outcomes Based on Particles Group Characteristics Using MLA Data and Machine Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Takeshi Shibuya,
  • Seyong Park,
  • Shun Ichige,
  • Taiga Sato,
  • Taisuke Sakakibara,
  • Tatsuhiro Ono,
  • Taro Kamiya,
  • Kazuya Sunada

摘要

This study aims to develop a support tool to beneficiation process for mineMine development projects by leveraging machine learningMachine learning on features of mineral particles obtained with electron microscopy and X-ray spectroscopy from a minimal quantity of ore samples. In this paper, a methodology to construct a predictive modelModel for flotationFlotation outcomes, including gradeGrade and recovery rateRecovery Rates under the same ore and flotationFlotation condition is proposed as a primary study. In the experiment, copper oreCopper ore feed subjected to various grindingGrinding conditions, along with the concentratesConcentrate and tailings obtained from flotationFlotation tests, were used. Training data for particle features were obtained using Mineral Liberation Analyzer and a predictive modelModel was subsequently constructed using machine learningMachine learning and evaluated for prediction accuracy. As a result, the predictions by the method were closer to the true values for test data under grindingGrinding conditions not included in the training dataset. This study contributes to solving the challenge of determining effective operating conditionsOperating conditions with a limited number of preliminary tests.