Search

CN-121998586-A - Intelligent identification green separation method and system for pegmatite type lithium ores

CN121998586ACN 121998586 ACN121998586 ACN 121998586ACN-121998586-A

Abstract

The invention discloses an intelligent identification green separation method and system for pegmatite lithium ores, which relate to the technical field of mineral identification and separation and comprise the following steps of constructing a mineral gene feature library; intelligent recognition and sorting decision model training; ore pretreatment and intelligent roughing waste disposal; the invention can realize high recovery rate of lithium minerals and high-grade concentrate output.

Inventors

  • WANG YU
  • ZHU HUAYUN
  • DENG YIPING
  • GUO JIANG
  • LUO XUFU
  • ZHAO CANYANG
  • WU YING
  • CHEN LONG
  • Liu Tilu

Assignees

  • 四川省非金属(盐业)地质调查研究所

Dates

Publication Date
20260508
Application Date
20260122

Claims (10)

  1. 1. The intelligent identification green separation method for the pegmatite type lithium ores is characterized by comprising the following steps of: s1, constructing a mineral gene feature library; s2, training an intelligent recognition and sorting decision model, namely training a mineral particle real-time recognition model and a sorting path decision model based on a mineral gene feature library; S3, ore pretreatment and intelligent roughing waste disposal, namely crushing raw ores, and identifying and separating low-value blocky materials by a mineral particle real-time identification model; S4, stage grinding and intelligent real-time sorting, namely stage grinding is carried out on the pre-selected ore, an intelligent real-time sorting unit is arranged after each stage of grinding, the on-line detection sensor detects that the data of the mineral particles after grinding are transmitted to the real-time identification model and the sorting path decision model, the particle identification and sorting decision is completed, and the particles are guided to a collecting channel through a programmable intelligent executing mechanism according to a decision instruction; S5, closed loop feedback and model optimization, namely performing online analysis on the collected mineral particles, comparing the actual sorting effect with the model prediction effect to form feedback data, and performing incremental learning or online optimization on a real-time identification model and a sorting path decision model of the mineral particles periodically or in real time by using the feedback data.
  2. 2. The intelligent recognition green separation method of the pegmatite type lithium ores is characterized in that in the step S1, core mineral gene characteristics are extracted, and a digital characteristic library is constructed.
  3. 3. The intelligent recognition green separation method of pegmatite type lithium ores according to claim 2, wherein the core mineral gene characteristics comprise mineralogy genes, physicochemical genes and response behavior genes.
  4. 4. The method for green sorting of intelligent identified peganite lithium ores according to claim 1, wherein in the step S2, a real-time identification model of mineral particles is input into a particle flow image or a spectrum sequence acquired by an online detection system, a sorting path decision model takes gene characteristics of particles identified by the real-time identification model of mineral particles as input, and a single or combined sorting method and operation parameters thereof are decided for the particles from a predefined sorting method library in combination with a preset target.
  5. 5. The method for green separation of intelligent identification pegmatite type lithium ores according to claim 1, wherein in the step S3, raw ores are crushed, screened and intelligent preselected firstly, and block ores after medium crushing or fine crushing are scanned and identified based on a real-time identification model of mineral particles, and lithium-rich ore blocks, lithium-poor or waste stone blocks which are dissociated by a monomer are separated, so that coarse-grain-level pre-waste discarding is realized.
  6. 6. The intelligent identification green separation method of the pegmatite type lithium ores, as set forth in claim 1, is characterized by further comprising the steps of S6, recycling and waste water purification, wherein the waste water generated in the separation process of S4 is subjected to coagulation, precipitation, adsorption and advanced oxidation treatment.
  7. 7. The intelligent identification green separation system for the peganite type lithium ores is characterized by comprising a mineral gene characteristic analysis platform, a model training and calculating server, an intelligent roughing waste throwing unit, a crushing module, a stage ore grinding loop, an intelligent real-time separation unit cluster, an on-line analysis and feedback module and a central control system, wherein the intelligent real-time separation unit cluster is connected in series, a discharge hole of the crushing module is connected with a feed hole of the intelligent roughing waste throwing unit, the stage ore grinding loop is connected with a discharge hole of the intelligent roughing waste throwing unit, and the intelligent real-time separation unit cluster comprises a plurality of intelligent real-time separation units which are arranged behind each ore grinding section of the stage ore grinding loop in series.
  8. 8. The intelligent-identification green separation system for pegmatite lithium ores, as set forth in claim 7, wherein the intelligent roughing waste disposal unit comprises a feeder, a sensor array and an intelligent executing mechanism.
  9. 9. The intelligent identified pegmatite lithium mine green separation system of claim 7, wherein the intelligent real-time separation unit comprises an online detection module, an intelligent decision module and a programmable separation execution module, and the intelligent decision module is communicated with a model training and calculating server.
  10. 10. The intelligent-identification green separation system for the pegmatite lithium ores, as set forth in claim 7, further comprises a tailings recycling and wastewater treatment module.

Description

Intelligent identification green separation method and system for pegmatite type lithium ores Technical Field The invention relates to the technical field of mineral identification and separation, in particular to an intelligent identification green separation method and system for pegmatite type lithium ores. Background The pegmatite type lithium ore is an important source of lithium resources, and the main lithium minerals are spodumene, lepidolite, petalite and the like. The traditional separation method mainly relies on an ore grinding-gravity separation-flotation process or an ore grinding-flotation process, and has the following inherent defects: 1. The process flow is complex, the efficiency is low, the multi-stage grinding and multi-stage separation are needed, the energy consumption is high, and the recovery rate of the lithium minerals, especially the ores with fine embedded granularity or complex symbiotic relation with gangue minerals, is not ideal. 2. The reagent consumption is large, the environment is not friendly, a large amount of regulator, collector and foaming agent are needed in the flotation process, water pollution and ecological damage can be caused, and the development requirements of the green mining industry are not met. 3. The separation precision depends on ore uniformity, the traditional process is sensitive to the fluctuation of raw ore properties, and when ore genes (such as mineral composition, embedding characteristics, crystallization degree and the like) are changed, the separation index is easy to fluctuate, and the process parameters are required to be frequently adjusted. 4. Coarse tailings are discharged, and resources are wasted, namely, in order to reduce overgrinding, the early-stage tailing discarding granularity is coarse, and the dissociated or intergrowth lithium minerals can be lost in the tailings. Disclosure of Invention The invention aims to overcome the defects of the prior art, and provides an intelligent identification peganite type lithium ore green sorting method and system, which are used for constructing a mineral gene feature library, establishing a mapping relation between mineral gene features and optimal sorting behaviors by using an intelligent model, and executing intelligent identification and sorting decisions in real time on a production line so as to realize high recovery rate and high-grade concentrate output of lithium minerals. The invention aims at realizing the following technical scheme that the intelligent identification green separation method for the pegmatite type lithium ores comprises the following steps: s1, constructing a mineral gene feature library, namely collecting representative ore samples aiming at target pegmatite type lithium ores. The sample is characterized by adopting a multi-dimensional analysis technology (such as MLA/QEMSCAN, laser induced breakdown spectroscopy LIBS, hyperspectral imaging, X-ray micro-area analysis and the like) system, and the following steps are extracted: Mineralogy genes include type, content, intercalation granularity, dissociation degree, symbiotic relation, crystal structure and crystal orientation of lithium mineral and main gangue mineral. Physicochemical genes, surface chemistry of the target mineral (such as zero electrical point, surface energy), characteristic spectrum (visible light-near infrared-short wave infrared, raman, LIBS spectrum), density, specific magnetization coefficient, conductivity, dielectric constant, and the like. Responding to behavior genes, namely, separating behavior difference data of minerals with different genetic characteristics in unit operations such as simulated or miniature reselection, magnetic separation, electric separation, flotation and the like. S2, training an intelligent recognition and sorting decision model, namely training the following intelligent model by utilizing a machine learning or deep learning algorithm based on the mineral gene feature library constructed in the step S1: The real-time identification model of mineral particles is input into a particle flow image or a spectrum sequence acquired by an on-line detection system (such as a hyperspectral camera, a LIBS probe and a high-speed vision system), and can accurately identify main lithium mineral types (such as spodumene and lepidolite), surface characteristics and continuous conditions of the particles or particle groups in real time. The model takes the identified particle gene characteristics (such as mineral type, continuous degree, surface pollution, granularity and the like) as input, and combines a preset optimization target (such as maximum recovery rate, maximum grade, minimum energy consumption, minimum medicament dosage or comprehensive benefit thereof), and dynamically decides the most effective single or combined sorting method and the optimal operation parameters (such as magnetic field strength, electric sorting voltage, flotation medicament dosage and dosage) for