WO2021196416A1 - 用于钨浮选精矿品位在线检测的光纤拉曼系统及方法 - Google Patents

用于钨浮选精矿品位在线检测的光纤拉曼系统及方法 Download PDF

Info

Publication number
WO2021196416A1
WO2021196416A1 PCT/CN2020/097212 CN2020097212W WO2021196416A1 WO 2021196416 A1 WO2021196416 A1 WO 2021196416A1 CN 2020097212 W CN2020097212 W CN 2020097212W WO 2021196416 A1 WO2021196416 A1 WO 2021196416A1
Authority
WO
WIPO (PCT)
Prior art keywords
tungsten
raman
spectrum
foam layer
flotation
Prior art date
Application number
PCT/CN2020/097212
Other languages
English (en)
French (fr)
Inventor
徐德刚
蔡耀仪
阳春华
桂卫华
Original Assignee
中南大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中南大学 filed Critical 中南大学
Publication of WO2021196416A1 publication Critical patent/WO2021196416A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/65—Raman scattering
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/01—Arrangements or apparatus for facilitating the optical investigation

Definitions

  • the invention relates to the technical field of online spectroscopy detection, in particular to an optical fiber Raman system and method for online detection of tungsten flotation concentrate grade.
  • Mineral flotation is the most widely used method for beneficiation of raw tungsten ore. It undergoes mineralization and foaming through an extremely complex physical and chemical reaction process.
  • the application of mineral flotation technology is to improve the grade of tungsten ore and meet the requirements of tungsten ore reduction smelting.
  • Concentrate grade of tungsten flotation is the content of useful substances in the beneficiation process, and is an important process indicator of tungsten flotation. Online real-time acquisition of tungsten flotation concentrate grade can guide the reasonable addition of reagents and the setting of working condition parameters in the flotation process, so that the flotation process is in the optimal production state.
  • Tungsten flotation produces various bubbles of different shapes and sizes to carry mineral particles. It is a three-phase complex body containing gas, liquid and solid, which has strong fluidity and is easy to rupture.
  • the grade of tungsten concentrate is directly related to the state of mineral foam, and it is also an important indicator of chemical dosing and liquid level control in the flotation process.
  • the existing tungsten concentrate grade analysis method can only obtain the detection value offline, that is, a small amount of foam is manually scooped and dried, and the mineral powder is chemically tested to obtain the concentrate grade data. This method is subject to human interference, complex testing process, and high cost.
  • the purpose of the present invention is to solve the problem that the grade of tungsten flotation concentrate is difficult to be detected online, and a fiber-optic Raman spectroscopy system for the online detection of tungsten flotation concentrate grade is designed. Method of concentrate grade.
  • An optical fiber Raman system for on-line detection of tungsten flotation concentrate grade comprising a laser generating device, a spectroscopic device and an industrial computer.
  • the laser generating device generates laser light and irradiates it vertically to the float in the tungsten flotation tank.
  • Select the foam layer to excite the minerals of the foam layer to generate Raman spectra the spectroscopic device collects the Raman spectra generated after the flotation foam layer is irradiated, and the industrial computer is communicatively connected to the spectroscopy device and receives the collected Raman spectra information.
  • the system, the laser generating device and the spectroscopic device emit laser light and collect the spectrum through an optical device
  • the optical device includes a composite Raman fiber probe and an optical collimating lens
  • the composite Raman fiber The probe is arranged above the flotation cell and faces the foam layer vertically, the optical collimating lens is fixed under the composite Raman fiber probe, and the laser generator emits toward the flotation foam layer through the composite Raman fiber probe Laser, the spectroscopic device collects Raman spectra through an optical collimator lens and a composite Raman fiber probe.
  • a method for on-line detection of tungsten flotation concentrate grade includes the following steps:
  • Step S1 Set up a system for collecting Raman spectra of the tungsten flotation foam layer
  • Step S2 Set a fixed spectrum integration time within a preset time interval, and obtain multiple Raman spectrum sampling results of the foam layer sample from this time interval, thereby forming a Raman spectrum set;
  • Step S3 Continue to collect the Raman spectrum of the foam layer sample for a period of time, construct multiple Raman spectrum sets based on step S2, and manually collect tungsten-selected foam layer samples corresponding to the Raman spectrum set, based on the ICP-MS method Obtain the grade value of its concentrate;
  • Step S4 Pre-processing the Raman spectra contained in each set to filter out the random noise of the Raman spectra and deduct the fluorescence scattering background signal;
  • Step S5 Calculate the average spectrum of the preprocessed Raman spectra in each set as a representative spectrum to eliminate detection errors caused by uneven distribution of minerals in the foam layer;
  • Step S6 Obtain the characteristic peak position information of the tungsten-containing mineral in the foam layer based on the molecular structure characteristics of the tungsten-containing mineral and its Raman spectrum characterization method;
  • Step S7 Combining the characteristic peak positions obtained in step S6, obtain the peak height, peak area and peak area information of the characteristic peaks of tungsten-containing minerals based on the Gaussian peak fitting method, so as to obtain the characteristic spectral bands of the tungsten-containing minerals ;
  • Step S8 Use the representative Raman spectrum of the foam layer sample as a training sample set, extract the characteristic spectrum information of the spectrum corresponding to each sample, and establish the foam layer Raman spectrum and the tungsten float based on the partial least squares regression method. Regression model between the grades of beneficiation concentrates;
  • Step S9 Preprocess the Raman spectrum of the foam layer collected online by tungsten flotation based on the step S3 to the step S7 to obtain a tungsten-containing mineral that filters out random noise, deducts the influence of fluorescence scattering background and redundant spectrum wavelength
  • the characteristic spectrum information of the Raman spectrum is then used as the input of the regression model established in step S8, so as to obtain the real-time concentrate grade of the tungsten flotation.
  • the concentrate grade value of the foam layer sample manually sampled in the step S3 is measured offline based on the ICP-MS method.
  • the preprocessing of the Raman spectrum of the foam layer mineral in the step S4 includes smoothing filtering, fluorescence background subtraction and spectral normalization.
  • the molecular structure feature of the tungsten-containing mineral and the Raman spectrum characterization method in the step S6 are obtained by matching the molecular bond of the tungsten-containing mineral with the characteristic peak of the Raman spectrum.
  • the Gaussian spectral peak fitting method in step S7 is realized based on the mathematical model of the Gaussian spectral peak, and the mathematical model formula is:
  • R is the Raman shift
  • I is the Raman scattering intensity under a certain Raman shift
  • I 0 is the baseline intensity.
  • S is a single Gaussian peak.
  • w is the half-height width of the Gaussian peak
  • R c is the Raman shift at the peak corresponding to the maximum spectral scattering intensity
  • ⁇ is the circumference of the circle. Therefore, a single Gaussian peak is composed of S, w and R c. Parameter decision.
  • the characteristic spectrum of the tungsten-containing mineral is obtained in step S7, and the characteristic spectrum of the spectrum is extracted based on the peak area and half-height of the characteristic peak in the Gaussian peak fitting result to determine its characteristic spectrum Segment range, that is, the range of Raman spectrum wavelength points contained in multiple characteristic peaks.
  • a regression model between the Raman spectrum of the foam layer and the grade of the tungsten flotation concentrate is established based on the partial least squares regression method, which is based on the partial least squares regression method on the tungsten flotation foam layer sample
  • the representative Raman spectroscopy of the data set after screening the characteristic spectrum is analyzed and processed to establish a quantitative model, including the following three steps:
  • the first step is to randomly select 70% of the total samples as the training set, and the remaining 30% as the test set, and divide it into two spectral data sets, the training sample set and the test sample set;
  • the partial least squares regression method is used to establish the corresponding regression model of the tungsten concentrate grade and the Raman spectrum;
  • the third step is to verify the established quantitative model through the test sample: Substitute the Raman spectrum data of the test sample into the quantitative model for prediction and fitting analysis, and judge the performance of the model by the error between the predicted value and the actual value. If the error between the two exceeds the error required by the concentration detection in the actual flotation process, the model parameters are adjusted by increasing the number of modeling samples and adjusting the number of principal components in the partial least squares model until The error meets the requirements, thereby optimizing the regression model.
  • the second step in the step S8, using the partial least squares regression method to establish the corresponding regression model of the tungsten concentrate grade and the Raman spectrum includes the following steps:
  • Step (1) First, perform principal component decomposition on the Raman spectrum matrix of the input training set samples, and then use the spectral principal component score matrix and concentrate grade vector to perform regression.
  • the formula for decomposing the spectral matrix X is as follows:
  • QT is the score matrix obtained by the decomposition of the spectral matrix X
  • P is the load matrix obtained by the decomposition of X
  • q k and p k are the k-th column vector of the matrix QT and P respectively
  • l is the number of columns of the above matrix
  • E X is the deviation produced by applying partial least squares to fit the matrix X
  • T is the transpose of the matrix
  • Step (2) Regression based on the scoring matrix QT obtained after the spectral matrix decomposition and the concentrate grade vector Y of the sample to obtain the correlation matrix B of the model is calculated as follows:
  • Step (3) Use the quantitative analysis model to predict the concentrate grade of the test set samples.
  • the specific step is to obtain the score matrix Q'according to the spectrum matrix X′ of the test set samples and the load matrix P of the model, and finally obtain it
  • the predicted value of concentrate grade is as follows:
  • the present invention has the characteristics of simple system hardware, high detection accuracy, and no need for sample preparation.
  • the optical fiber Raman spectroscopy system designed in the present invention can collect the Raman spectra generated by the tungsten selection groove foam layer minerals in real time online in an open environment through a collimator lens, a long-distance laser and a collection optical fiber.
  • the special hardware structure design of the system can weaken the influence of strong on-site vibration, acid mist corrosion of the flotation cell and electromagnetic interference on the instrument to a certain extent, and ensure the accuracy and reliability of spectrum acquisition.
  • the present invention accurately obtains the representative spectra of each concentrate grade, reduces the influence of uneven mineral distribution on the accuracy of the quantitative prediction model, and uses smoothing filtering, fluorescence background subtraction algorithms and based on The characteristic spectral band screening of Gaussian peak fitting can eliminate the interference of random noise, fluorescence scattering signal and non-information wavelength, and improve the representativeness of the spectral signal.
  • Figure 1 is a schematic diagram of the hardware structure of the fiber-optic Raman system for on-line detection of tungsten flotation concentrate grade
  • Figure 2 is a flowchart of the establishment of a concentrate grade prediction model
  • Figure 3 is a block diagram of the whole process of online detection of the location of minerals in the foam layer of the tungsten flotation selection tank;
  • 1 is a laser generator with a power of 500mW and a laser wavelength of 785nm
  • 2 is a miniature refrigerated Raman spectrometer
  • 3 is a laser incident fiber
  • 4 is a collection fiber
  • 5 is a composite spectroscopic fiber probe
  • 6 is a focusing Lens
  • 7 is the foam layer of the tungsten selection tank
  • 8 is the surface layer of the overflow tank
  • 9 is the tungsten selection tank
  • 10 is the tungsten selection overflow tank
  • 11 is the Raman spectroscopy system and industrial computer connection data fiber
  • 12 is the industrial Computer
  • 13 is the outer protective cover of the spectrum probe fixing bracket
  • 14 is the spectrum probe fixing bracket
  • 15 is the spectrum probe fixing base.
  • the hardware structure of the fiber-optic Raman system for online detection of tungsten flotation concentrate grade is shown in Figure 1. It is mainly composed of a 785nm laser generator 1 with a power of 500mW, a miniature refrigerated Raman spectrometer 2, a laser incident fiber 3, a collection fiber 4, and a composite Raman spectroscopy fiber probe 5, optical collimation lens 6, Raman spectroscopy system and industrial computer connection data fiber 11, industrial computer 12, spectrum probe fixing bracket outer protective cover 13, spectrum probe fixing bracket 14 and spectrum probe fixing base 15 composition.
  • the laser generator 1 is used to excite a sample of the foam layer of the flotation cell to generate a Raman spectrum.
  • the miniature refrigerated Raman spectrometer 2 is used to collect the Raman scattering of a sample of the tungsten-selected foam layer.
  • the composite Raman spectroscopy optical fiber probe 5 can be reused for conducting laser light and collecting spectra.
  • the optical collimator lens is used to increase the laser focus distance, thereby increasing the Raman spectrum detection distance.
  • the industrial computer 12 is used for online construction of a quantitative model of the tungsten concentrate grade spectrum.
  • the outer protective cover 13 of the spectrum probe fixing bracket, the spectrum probe fixing bracket 14 and the spectrum probe fixing base 15 are used to fix and protect the composite Raman spectroscopy optical fiber probe 5 in combination.
  • the tungsten beneficiation tank 9 is the place where the mixing reaction of the slurry and the drug produced by the rough separation of tungsten occurs, so that the mineral binding foam floats to the surface of the slurry and forms a mineralized foam. Its surface is the area where the most foam is produced, which has a good Isolation.
  • the tungsten beneficiation overflow tank 10 is a vessel for holding the mineralized foam scraped from the surface of the flotation tank by the scraper, which can further improve the mineral content and provide raw materials for the final smelting of tungsten ingots.
  • the composite spectroscopic probe of the optical fiber Raman system is connected to the optical collimating lens 6, and the fixed bracket 14 makes it perpendicular to the surface of the tungsten finely selected foam layer.
  • the 500mw 785nm laser generator continuously generates stable parallel laser light which is conducted by the incident optical fiber 3 to the compound spectroscopic probe 5, and is further focused by the focusing lens and injected into the tungsten selected foam layer 7 with tungsten mineral particles adhered to the surface of the foam layer. Finally, the tungsten-containing minerals are excited to produce Raman scattering, which is transmitted to the miniature refrigerated Raman spectrometer 2 by the collecting optical fiber 4 in the composite spectroscopic probe 5, and finally the Raman spectrum collected by the spectrometer is sent to the industrial computer 12.
  • the mass percentage of tungsten-containing minerals in the foam layer of the tungsten beneficiation tank is also called tungsten flotation concentrate grade.
  • the online detection of tungsten flotation concentrate grade through the above system includes the following steps:
  • the refrigerated miniature Raman spectrometer in the optical fiber Raman spectroscopy system is placed in the industrial control room of the tungsten flotation site, and at the same time, it is connected to the composite Raman spectroscopy probe above the tungsten selection tank through a high-performance optical fiber. Collect the sample of the foam layer of the tungsten flotation process and its Raman spectrum in real time.
  • the integration time of a single spectrum in this embodiment is 10s. In actual implementation, the spectrum integration time The spectrum acquisition time of the Raman spectrometer is set and changed.
  • the laser wavelength used in the Raman fiber spectroscopy system is 785nm
  • the laser output power is 100mW to 500mW
  • the spectral wavelength ranges from 150cm -1 to 2000cm -1
  • the spectral resolution is greater than 10cm -1 .
  • the Raman spectrum of each sample is converted by spectrometer analog-to-digital conversion and output to an industrial computer for storage to form a Raman spectrum library of tungsten-selected foam layer samples.
  • the Raman spectrum is smoothed by the S-G filtering method.
  • the S-G smoothing filter is based on the polynomial least square fitting method to fit the data in the smoothing window. The specific method is as follows.
  • x be the Raman spectrum signal of the collected sample of the tungsten-selected foam layer
  • z is the smoothed signal
  • n its wave number range
  • the unsmoothness of z can be expressed by the sum of squares of its difference, as shown in formula (2).
  • the constructor describes the weighted sum between fidelity and unsmoothness as shown in formula (3).
  • Dz is the result of the matrix z differentiation, and the penalty factor ⁇ directly adjusts the smoothness of the fitted data, and the least square method is applied to the function Q to obtain the smoothed Raman spectrum.
  • the baseline is fitted based on the airPLS method, that is, by adjusting the penalty factor and least squares iteration, changing the square sum weight of the error between the fitted baseline and the original spectral signal, so as to gradually approach The actual baseline shape, and then deduct the influence of the fluorescence scattering signal.
  • the above determination of the position of the characteristic spectral peak is realized based on the molecular structure characteristics of the tungsten-containing mineral.
  • the Gaussian spectral peak fitting method is used to obtain the Gaussian mathematical model of the characteristic spectral peak, and then the characteristic spectral peak width is obtained.
  • the spectral characteristic peaks of tungsten-containing minerals are determined by analyzing the molecular structure and chemical bonds of the minerals, thereby determining the molecular vibration mode, and finally obtaining the spectral characteristic peak information.
  • the Raman spectrum Gaussian peak fitting method is realized by the determined peak position of the tungsten-containing mineral spectrum. For a single Raman spectrum collected, the Raman spectrum curve to be fitted is composed of multiple spectral characteristic peaks, which can be expressed as:
  • i is the index value of the Gaussian peak
  • M is the number of Raman characteristic peaks of tungsten-containing minerals
  • R j is the Raman shift at the jth sampling point on the Raman spectrum
  • S i , ⁇ i and R ci are the peak area, half-width and peak position of the i-th Gaussian fitted spectrum peak, respectively.
  • the Gaussian spectral peak fitting process is the process of optimizing the solution of the Gaussian spectral peak equation. It is usually based on the method of minimizing the mean square error to make the fitting curve as close to the original spectral curve as possible, so as to solve the optimal S i and ⁇ i .
  • the process can be described as follows:
  • the characteristic spectral peak area and half-height width are determined, so as to accurately screen the characteristic spectral bands and reduce the interference of the wavelength points of the non-information Raman spectrum.
  • tungsten concentrate grade modeling sample library based on the preprocessed Raman spectra of multiple tungsten concentrate samples and their concentrate grade values, and use multivariate statistical analysis methods to analyze and process Raman spectra data to establish and establish tungsten concentrate grades Quantitative analysis model.
  • the grade value of the concentrate is obtained based on step 2, and the multivariate statistical analysis method used is partial least squares regression.
  • the purpose is to obtain a tungsten concentrate grade prediction model, which can be used for online prediction of tungsten flotation concentrate grade.
  • a quantitative analysis model of tungsten concentrate grade is established based on the method shown in Figure 3. The specific method is as follows:
  • the partial least squares (PLSR) modeling method eliminates the noise data and obtains an accurate quantitative prediction model of concentrate grade.
  • Partial least squares (PLSR) modeling method is used to decompose the spectral array X, and at the same time, a quantitative model of concentrate grade is obtained by using concentrate grade vector regression.
  • the decomposed matrix T is the score matrix
  • P is the load matrix obtained by the decomposition of X
  • t k and p k are the k- th column vector of the matrix T and P respectively
  • l is the number of columns of the above matrix
  • E X is To apply partial least squares method to fit the deviation of matrix X.
  • the number of principal components of the spectrum matrix X is determined based on the indicator function method (IND), that is, the number of score vectors contained in the score vector matrix T, as shown in the following formula.
  • IND indicator function method
  • the calculation method of the least square solution B of the regression coefficient b is as shown in the following formula.
  • the established tungsten selection tank foam layer ore position prediction model can be used to obtain the concentrate grade value online in the tungsten flotation process.
  • the specific steps are as follows:
  • Step (1) Collect the real-time Raman spectrum of the foam layer of the tungsten selection tank online and complete the preprocessing of the spectrum to obtain the spectrum array X.

Landscapes

  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
  • Spectrometry And Color Measurement (AREA)

Abstract

一种用于钨浮选精矿品位在线检测的光纤拉曼系统及方法。基于拉曼光谱技术，设计了钨浮选中浮选槽泡沫层拉曼光谱在线测量装置，构建了泡沫层矿物的拉曼光谱数据在线处理系统，从而提取实时采集的钨浮选泡沫层拉曼光谱相关特征并构建精矿品位在线定量计算模型，最终在线获取钨浮选精矿品位结果。光纤拉曼系统基于不同钨精矿品位下的拉曼光谱特征向量建立钨精矿定量分析模型，在浮选过程中实时获取泡沫层中的钨精矿品位值，能够在线监测钨浮选中重要生产和考核指标，为钨浮选全流程控制提供了重要参数。

Description

用于钨浮选精矿品位在线检测的光纤拉曼系统及方法 技术领域
本发明涉及在线光谱检测技术领域，具体涉及一种用于钨浮选精矿品位在线检测的光纤拉曼系统及方法。
背景技术
矿物浮选是钨原矿应用最广泛的一种选矿方法，通过极其复杂的物理化学反应过程进行矿化起泡。应用矿物浮选技术是为了提高钨原矿品位，达到钨矿还原冶炼的要求。钨浮选中精矿品位是精选工序中有用物质的含量，是钨浮选中重要的工艺指标。在线实时获取钨浮选精矿品位能够指导浮选过程中药剂的合理添加和工况参数的设置，使得浮选过程处于最优生产状态。
钨浮选中产生形态和大小不等的各种气泡来携带矿粒，是一种包含有气、液、固的三相复杂体，其流动性强且容易破裂。钨精矿品位与矿物泡沫状态直接相关，同时也是浮选过程中加药和液位控制的重要指标。现有的钨精矿品位分析方法只能离线获取检测值，即人工舀取少量泡沫并晒干，将矿物粉末应用化学方法化验得到精矿品位数据。这种方法受人为干扰较大、化验过程复杂、成本较高，往往一天只能化验一份样本，难以实时指导钨浮选中的加药和工况参数调整，直接影响最终钨精矿的回收率。少数浮选企业应用X射线荧光光谱技术直接测量浮选槽泡沫的矿物品位，但对待检测泡沫样品的制备要求较高，且由于仪器会产生X射线，从而检测过程中具有较强放射性影响，对检测环境的密封性和使用者的安全操作要求较高，提高了单次检测成本和复杂度。拉曼光谱是一种对样本制备要求较低的非接触式光谱检测技 术，获取光谱速度快，且不受水的影响，从而能够直接检测精选过程中泡沫层钨矿品位值，具有实现钨浮选精矿品位在线检测的潜力。
发明内容
本发明的目的在于解决钨浮选中精矿品位难以在线检测的问题，设计了一种用于钨浮选精矿品位在线检测的光纤拉曼光谱系统，同时针对系统给出了其在线检测钨精矿品位的方法。
本发明的技术方案是，
一种用于钨浮选精矿品位在线检测的光纤拉曼系统，包括激光发生装置、光谱装置和工业计算机，所述的激光发生装置产生激光并竖直的照射至钨浮选槽内的浮选泡沫层上以激发泡沫层矿物产生拉曼光谱，所述的光谱装置采集浮选泡沫层被照射后产生的拉曼光谱，所述的工业计算机通信连接至光谱装置并接收采集的拉曼光谱信息。
所述的系统，所述的激光发生装置和光谱装置通过光学装置来发射激光和收集光谱，所述的光学装置包括复合型拉曼光纤探头和光学准直透镜，所述的复合型拉曼光纤探头设置于浮选槽上方并竖直朝向泡沫层，所述的光学准直透镜固定于复合型拉曼光纤探头下，所述的激光发生装置通过复合型拉曼光纤探头朝向浮选泡沫层发射激光，所述的光谱装置通过光学准直透镜和复合型拉曼光纤探头采集拉曼光谱。
一种用于钨浮选精矿品位在线检测的方法，包括以下步骤：
步骤S1：设置采集钨浮选泡沫层拉曼光谱的系统；
步骤S2：在预设时间区间内，设定固定的光谱积分时间，从此时间区间内获得泡沫层样本的多个拉曼光谱采样结果，从而构成拉曼光谱集合；
步骤S3：持续采集一段时间内泡沫层样本的拉曼光谱，基于步骤S2构建 多个拉曼光谱集合，同时人工采集与拉曼光谱集合相对应的钨精选泡沫层样本，基于ICP-MS方法获取其精矿品位值；
步骤S4:对各个集合中所包含的拉曼光谱进行预处理，滤除拉曼光谱的随机噪声及扣除其荧光散射背景信号；
步骤S5：对各个集合中预处理后的拉曼光谱求取其平均光谱作为代表性光谱，消除泡沫层矿物分布不均带来的检测误差；
步骤S6：基于含钨矿物的分子结构特征及其拉曼光谱表征方法获取泡沫层中含钨矿物的特征峰峰位信息；
步骤S7：结合步骤S6中所获取的特征峰峰位，基于高斯谱峰拟合方法获取含钨矿物特征谱峰的峰高、峰面积和谱峰区域信息，从而获取含钨矿物的特征谱段；
步骤S8：利用所述泡沫层样本的代表性拉曼光谱作为训练样本集合，提取其每个样本所对应光谱的特征谱段信息，基于偏最小二乘回归方法建立泡沫层拉曼光谱与钨浮选精矿品位间的回归模型；
步骤S9：对钨浮选中在线采集的泡沫层拉曼光谱基于所述步骤S3至所述步骤S7进行预处理，得到滤除随机噪声、扣除荧光散射背景和冗余光谱波长影响的含钨矿物拉曼光谱的特征谱段信息，其后将该特征谱段信息作为所述步骤S8中所建立回归模型的输入，从而获取钨浮选中的实时精矿品位。
所述的方法，所述步骤S3中人工采样的泡沫层样本的精矿品位值基于ICP-MS方法离线测定。
所述的方法，所述步骤S4中对泡沫层矿物拉曼光谱的预处理包括平滑滤波、荧光背景扣除和光谱归一化。
所述的方法，所述步骤S6中含钨矿物的分子结构特征及其拉曼光谱表征 方法是通过匹配含钨矿物分子键与拉曼光谱特征峰获取。
所述的方法，所述步骤S7中高斯谱峰拟合方法是基于高斯谱峰的数学模型实现，数学模型公式为：
Figure PCTCN2020097212-appb-000001
其中，R是拉曼位移，I是一定拉曼位移下的拉曼散射强度，I 0为基线强度，当原始拉曼光谱经过基线校正后，I 0的值为0，S为单个高斯谱峰的峰面积，w为高斯谱峰的半高宽，R c为最大光谱散射强度所对应的谱峰处的拉曼位移，π为圆周率，因此，单个高斯峰由S、w和R c三个参数决定。
所述的方法，所述步骤S7中获取含钨矿物的特征谱段，是提取光谱的特征谱段基于高斯分峰拟合结果中特征谱峰的峰面积和半高宽，来确定其特征谱段范围，即多个特征谱峰所包含的拉曼光谱波长点的范围。
所述的方法，所述步骤S8中基于偏最小二乘回归方法建立泡沫层拉曼光谱与钨浮选精矿品位间的回归模型，是基于偏最小二乘回归方法对钨浮选泡沫层样本的代表性拉曼光谱筛选特征谱段后的数据集合进行分析处理从而建立定量模型，包含以下三个步骤：
第一步，随机选取总样本中的70％作为训练集，而剩余30％样本作为测试集，将其划分成训练样本集和测试样本集两个光谱数据集；
第二步，基于训练样本集，采用偏最小二乘回归法建立钨精矿品位和拉曼光谱的对应回归模型；
第三步，通过测试样本对已建立的定量模型进行验证：将测试样本的拉曼光谱数据代入定量模型中进行预测和拟合分析，通过预测值与实际值之间的误差来判断模型性能，若两者之间的误差超过实际浮选过程中精矿品位检 测所要求的误差，则通过增大建模样本数、调整偏最小二乘模型中的主成分个数的方式调整模型参数，直至误差符合要求，从而优化回归模型。
所述的方法，所述步骤S8中第二步，采用偏最小二乘回归法建立钨精矿品位和拉曼光谱的对应回归模型包括如下步骤：
步骤(1)：首先对输入的训练集样本的拉曼光谱矩阵进行主成分分解，其后利用光谱主成分得分矩阵和精矿品位向量进行回归，其中对光谱矩阵X进行分解的公式如下：
Figure PCTCN2020097212-appb-000002
公式中，QT为光谱矩阵X分解得到的得分矩阵，而P则为X分解得到的载荷矩阵，q k和p k分别为矩阵QT和P的第k列列向量，l为上述矩阵的列数，而E X则为应用偏最小二乘法拟合矩阵X所产生的偏差，矩阵符号T为求矩阵的转置；
步骤(2)：基于光谱矩阵分解后所得到的得分矩阵QT及样本的精矿品位向量Y进行回归，从而得到模型的关联矩阵B计算如下：
B＝(Q TQ) -1Q TY
步骤(3)：利用定量分析模型对测试集样本的精矿品位进行预测，具体步骤为根据测试集样本的光谱矩阵X′以及模型的载荷矩阵P求出其得分矩阵Q′，从而最终得到其精矿品位预测值如下：
y＝(X′) TB′
其中，B′＝[(Q′) T(Q′)] -1(Q′) TY。
与现有技术相比，本发明的技术效果如下：
1)本发明通过所设计的光纤拉曼光谱系统能够在钨浮选中在线获取不 同精矿品位泡沫层样本的拉曼光谱并建立定量校正模型，从而在线实时检测钨精矿品位，有效解决了钨浮选中精矿品位难以在线测定的问题，其测量精度能够达到工业现场要求。
2)本发明具有系统硬件简单、检测精度高、无需样本制备的特点。本发明所设计的光纤拉曼光谱系统通过准直透镜及长距离激光和收集光纤能够在开放环境下在线实时收集钨精选槽泡沫层矿物所产生的拉曼光谱。
3)系统特殊的硬件结构设计能够在一定程度上削弱现场强烈震动、浮选槽酸雾腐蚀和电磁干扰对仪器的影响，保证了光谱获取的准确和可靠性。
4)本发明针对多次测量并平均的方法准确得到各个精矿品位下的代表光谱，减小了矿物分布不均匀对定量预测模型精度的影响，所采用的平滑滤波、荧光背景扣除算法及基于高斯谱峰拟合的特征谱段筛选能够消除随机噪声、荧光散射信号和无信息波长的干扰，提高了光谱信号的代表性。
附图说明
图1是钨浮选精矿品位在线检测光纤拉曼系统的硬件结构示意图；
图2是精矿品位预测模型建立的流程图；
图3是在线检测钨浮选中精选槽泡沫层矿物品位的全流程框图；
图1中，1为功率为500mW，激光波长为785nm的激光发生器，2为微型制冷式拉曼光谱仪，3为激光入射光纤，4为收集光纤，5为复合型光谱光纤探头，6为聚焦透镜，7为钨精选槽泡沫层，8为溢流槽表层，9为钨精选槽，10为钨精选溢流槽，11为拉曼光谱系统与工业计算机连接数据光纤，12 为工业计算机，13为光谱探头固定支架外保护套、14为光谱探头固定支架、15为光谱探头固定底座。
具体实施方法
下面结合附图和具体实施过程对本发明进行详细说明，所举实例只用于解释本发明，并非限定本发明范围。
钨浮选精矿品位在线检测光纤拉曼系统的硬件结构如图1所示，主要由功率500mW的785nm激光发生器1，微型制冷式拉曼光谱仪2，激光入射光纤3，收集光纤4，复合型拉曼光谱光纤探头5，光学准直透镜6，拉曼光谱系统与工业计算机连接数据光纤11，工业计算机12，光谱探头固定支架外保护套13、光谱探头固定支架14和光谱探头固定底座15组成。激光发生器1用于激发浮选槽泡沫层样本产生拉曼光谱。微型制冷式拉曼光谱仪2用于收集钨精选泡沫层样本拉曼散射。复合型拉曼光谱光纤探头5能够复用于传导激光和收集光谱。光学准直透镜用于实现激光聚焦距离的提升，从而提升拉曼光谱检测距离。工业计算机12用于在线构建钨精矿品位光谱定量模型。光谱探头固定支架外保护套13、光谱探头固定支架14和光谱探头固定底座15用于组合起来对复合型拉曼光谱光纤探头5进行固定及保护。钨精选槽9是钨粗选产生的矿浆与药物混合反应发生的场所，使得矿物粘合泡沫之上，浮到矿浆面并形成矿化泡沫，其表面是泡沫产生最多的区域，具有良好的隔离性。钨精选溢流槽10是盛放由刮板在浮选槽表面刮出的矿化泡沫的器皿，能够进一步提高矿物品位，为最终钨锭的冶炼提供原材料。光纤拉曼系统的复合型光谱探头连接着光学准直透镜6，由固定支架14使其垂直于钨精选泡沫层表面。500mw的785nm激光发生器持续产生稳定的平行激光由入射光纤3传导至复合型光谱探头5中并通过聚焦透镜进一步聚焦并射入钨精选泡沫层7 表面粘附钨矿物颗粒的泡沫层中，最终激发含钨矿物产生拉曼散射并由复合光谱探头5中的收集光纤4传导入微型制冷式拉曼光谱仪2，并最终将光谱仪收集到的拉曼光谱发送至工业计算机12。
在浮选领域，钨精选槽泡沫层中含钨矿物的质量百分比也称为钨浮选精矿品位，本实施例通过以上系统实现钨浮选精矿品位在线检测包含以下步骤：
将光谱探头垂直于钨精选槽上方，探头与浮选槽泡沫层距离为15cm左右。同时，将光纤拉曼光谱系统中的制冷微型拉曼光谱仪放置于钨浮选现场的工控室，同时通过高性能光纤与钨精选槽上方的复合拉曼光谱探头相连接。实时采集钨浮选过程精选槽泡沫层样本和其拉曼光谱若干份，在采集拉曼光谱信号时，本实施例的单个光谱的积分时间为10s，实际实施中，光谱积分时间通过对制冷型拉曼光谱仪的光谱采集时间进行设置而改变。对连续采集到的三个拉曼光谱求取平均值，并将其作为钨精选的表征光谱。拉曼光纤光谱系统所采用的激光器波长为785nm，激光输出功率为100mW至500mW，光谱波长范围从150cm -1～2000cm -1，光谱分辨率大于10cm -1。
对于采集的样本进行取样、烘干。基于XRF荧光光谱仪，采用标准XRF定量分析来离线确定各个样本的精矿品位精确值，从而作为钨浮选精矿品位的实时标定值，将其作为初始建模样本库。
将各个样本的拉曼光谱经过光谱仪模数转换并输出到工业计算机中存储，构成钨精选泡沫层样本的拉曼光谱库。
在测量过程中，由于拉曼散射信号较弱、仪器本身存在设计缺陷和采集过程中外界干扰的影响，从而采集到的拉曼光谱会叠加噪声信号，而噪声是一种无用信息，通常对有用信息的提取造成影响。为了消除噪声的影响，采用基于S-G滤波法对拉曼光谱进行平滑处理，S-G平滑滤波是基于多项式最 小二乘拟合方法对平滑窗口内数据进行拟合，具体方法如下。
设x为采集到的钨精选泡沫层样本的拉曼光谱信号，z为平滑后的信号，其波数范围均为n，z相对于x的失真程度可以用二者之间的误差平方和表示，如公式(1)所示。
Figure PCTCN2020097212-appb-000003
z的不平滑度可以用其差分平方和表示，如公式(2)所示。
Figure PCTCN2020097212-appb-000004
引入惩罚因子，构造函数描述保真度与不平滑度之间的加权和如公式(3)所示。
Q＝F+λR＝||x-z|| 2+λ||Dz|| 2  (3)
公式(3)中，Dz即为矩阵z微分的结果，惩罚因子λ直接调节拟合数据的平滑程度，对函数Q应用最小二乘法，即可得到平滑后的拉曼光谱。
测量得到的拉曼光谱中会存在强荧光散射信号的干扰，造成整个光谱基线发生漂移，严重影响定量建模的准确性。为了校正光谱基线和扣除荧光背景的影响，基于airPLS方法对基线进行拟合，即通过调节惩罚因子和最小二乘迭代，改变拟合基线和原始光谱信号之间误差的平方和权重，从而逐步接近实际的基线形状，进而扣除荧光散射信号的影响。
获取含钨矿物的分子结构特征及其拉曼光谱表征结果，其拉曼光谱如图2所示。可以发现含钨矿物的拉曼光谱特征谱峰大多分布于1200-2000cm -1范围内，因此可以初步筛选出1000-2500cm -1作为含钨矿物的关键光谱信息区域。同时，基于含钨矿物的分子结构特征确定其特征谱峰位置，可知其特征谱峰主要位于1268cm -1,1337cm -1,1434cm -1,1499cm -1和1800cm -1。上述确定其特征 谱峰位置是基于含钨矿物的分子结构特征来实现的，同时采用高斯谱峰拟合方法获取其特征谱峰的高斯数学模型，进而得到其特征谱峰宽度。具体来说，含钨矿物光谱特征峰是通过分析其矿物分子结构及化学键，从而确定其分子振动模式，最终得到其光谱特征峰信息。拉曼光谱高斯谱峰拟合方法通过已确定的含钨矿物光谱特征谱峰位置来实现。针对采集到的单个拉曼光谱而言，要拟合的拉曼光谱曲线由多个光谱特征峰组成，可以表示为：
Figure PCTCN2020097212-appb-000005
其中，i为高斯谱峰的索引值，M为含钨矿物的拉曼特征谱峰个数，R j为拉曼光谱上第j个采样点处的拉曼位移，
Figure PCTCN2020097212-appb-000006
为拟合得到的拉曼光谱，S i，ω i和R ci分别为第i个高斯拟合谱峰的峰面积、半宽高和峰位置。
高斯谱峰拟合过程就是对高斯谱峰方程的优化求解过程，通常基于最小化均方误差方法使拟合曲线尽可能逼近原始光谱曲线，从而求解最优的S i和ω i，其寻优过程可以描述如下：
Figure PCTCN2020097212-appb-000007
s.t.ω i>0,S i>0
基于高斯谱峰拟合方法确定其特征谱峰面积及半高宽，从而精确筛选其特征谱段，并减少无信息拉曼光谱波长点的干扰。
基于多个钨精矿样本预处理后的拉曼光谱及其精矿品位值构建钨精矿品位建模样本库，采用多元统计分析方法对拉曼光谱数据进行分析处理建立并建立钨精矿品位定量分析模型。其精矿品位值基于步骤二获取，所采用的的多元统计分析方法为偏最小二乘回归法。目的是获取钨精矿品位预测模型， 从而用于钨浮选精矿品位的在线预测。基于图3所示的方法建立钨精矿品位定量分析模型，具体方法如下：
不同矿物品位拉曼代表光谱中存在无用的噪声信息，影响定量建模的精度，从而偏最小二乘(PLSR)建模方法消除消除噪声数据并得到精确的精矿品位定量预测模型。偏最小二乘(PLSR)建模方法对光谱阵X进行矩阵分解，同时利用精矿品位向量回归得到精矿品位定量模型。
首先，对光谱矩阵X进行分解，如下式所示。
Figure PCTCN2020097212-appb-000008
分解后的矩阵T为得分矩阵，而P则为X分解得到的载荷矩阵，t k和p k分别为矩阵T和P的第k列列向量，l为上述矩阵的列数，而E X则为应用偏最小二乘法拟合矩阵X所产生的偏差。
基于指示函数法(IND)确定光谱阵X的主成分数，即得分向量阵T所含得分向量个数，如下式所示。
Figure PCTCN2020097212-appb-000009
逐一计算f＝1开始不同f取值下的H IND值，选取H IND极小值时的f即为主成分数，并获得其对应的f个得分向量矩阵T＝[t 1,t 2,…,t f]与矿物品位对应的浓度向量y进行多元线性回归(MLR)，即得到其偏最小二乘回归(PLSR)模型如下式所示。
Y＝Tb+E
回归系数b的最小二乘解B的计算方法如下式所示。
B＝(T TT) -1T TY
所建立的钨精选槽泡沫层矿物品位预测模型可用于钨浮选过程中在线获取精矿品位值，具体步骤如下：
步骤(1)：在线采集钨精选槽泡沫层的实时拉曼光谱并完成对光谱的预处理，得到光谱阵X。
步骤(2)：对光谱阵X分解得到载荷矩阵P，基于指示函数法确定其主成分数f，并计算其前f个得分向量矩阵T＝[t 1,t 2,…,t f]。
步骤(3)：基于已经建立的钨精选槽泡沫层矿物品位预测模型回归系数B及拟合残差E，计算得到实时矿物品位y＝TB+E。

Claims (10)

  1. 一种用于钨浮选精矿品位在线检测的光纤拉曼系统，其特征在于，包括激光发生装置、光谱装置和工业计算机，所述的激光发生装置产生激光并竖直的照射至钨浮选槽内的浮选泡沫层上以激发泡沫层矿物产生拉曼光谱，所述的光谱装置采集浮选泡沫层被照射后产生的拉曼光谱，所述的工业计算机通信连接至光谱装置并接收采集的拉曼光谱信息。
  2. 根据权利要求1所述的系统，其特征在于，所述的激光发生装置和光谱装置通过光学装置来发射激光和收集光谱，所述的光学装置包括复合型拉曼光纤探头和光学准直透镜，所述的复合型拉曼光纤探头设置于浮选槽上方并竖直朝向泡沫层，所述的光学准直透镜固定于复合型拉曼光纤探头下，所述的激光发生装置通过复合型拉曼光纤探头朝向浮选泡沫层发射激光，所述的光谱装置通过光学准直透镜和复合型拉曼光纤探头采集拉曼光谱。
  3. 一种用于钨浮选精矿品位在线检测的方法，其特征在于，包括以下步骤：
    步骤S1：设置采集钨浮选泡沫层拉曼光谱的系统；
    步骤S2：在预设时间区间内，设定固定的光谱积分时间，从此时间区间内获得泡沫层样本的多个拉曼光谱采样结果，从而构成拉曼光谱集合；
    步骤S3：持续采集一段时间内泡沫层样本的拉曼光谱，基于步骤S2构建多个拉曼光谱集合，同时人工采集与拉曼光谱集合相对应的钨精选泡沫层样本，基于ICP-MS方法获取其精矿品位值；
    步骤S4:对各个集合中所包含的拉曼光谱进行预处理，滤除拉曼光谱的随机噪声及扣除其荧光散射背景信号；
    步骤S5：对各个集合中预处理后的拉曼光谱求取其平均光谱作为代表性 光谱，消除泡沫层矿物分布不均带来的检测误差；
    步骤S6：基于含钨矿物的分子结构特征及其拉曼光谱表征方法获取泡沫层中含钨矿物的特征峰峰位信息；
    步骤S7：结合步骤S6中所获取的特征峰峰位，基于高斯谱峰拟合方法获取含钨矿物特征谱峰的峰高、峰面积和谱峰区域信息，从而获取含钨矿物的特征谱段；
    步骤S8：利用所述泡沫层样本的代表性拉曼光谱作为训练样本集合，提取其每个样本所对应光谱的特征谱段信息，基于偏最小二乘回归方法建立泡沫层拉曼光谱与钨浮选精矿品位间的回归模型；
    步骤S9：对钨浮选中在线采集的泡沫层拉曼光谱基于所述步骤S3至所述步骤S7进行预处理，得到滤除随机噪声、扣除荧光散射背景和冗余光谱波长影响的含钨矿物拉曼光谱的特征谱段信息，其后将该特征谱段信息作为所述步骤S8中所建立回归模型的输入，从而获取钨浮选中的实时精矿品位。
  4. 根据权利要求3所述的方法，其特征在于，所述步骤S3中人工采样的泡沫层样本的精矿品位值基于ICP-MS方法离线测定。
  5. 根据权利要求3所述的方法，其特征在于，所述步骤S4中对泡沫层矿物拉曼光谱的预处理包括平滑滤波、荧光背景扣除和光谱归一化。
  6. 根据权利要求3所述的方法，其特征在于，所述步骤S6中含钨矿物的分子结构特征及其拉曼光谱表征方法是通过匹配含钨矿物分子键与拉曼光谱特征峰获取。
  7. 根据权利要求3所述的方法，其特征在于，所述步骤S7中高斯谱峰拟合方法是基于高斯谱峰的数学模型实现，数学模型公式为：
    Figure PCTCN2020097212-appb-100001
    其中，R是拉曼位移，I是一定拉曼位移下的拉曼散射强度，I 0为基线强度，当原始拉曼光谱经过基线校正后，I 0的值为0，S为单个高斯谱峰的峰面积，w为高斯谱峰的半高宽，R c为最大光谱散射强度所对应的谱峰处的拉曼位移，π为圆周率，因此，单个高斯峰由S、w和R c三个参数决定。
  8. 根据权利要求3所述的方法，其特征在于，所述步骤S7中获取含钨矿物的特征谱段，是提取光谱的特征谱段基于高斯分峰拟合结果中特征谱峰的峰面积和半高宽，来确定其特征谱段范围，即多个特征谱峰所包含的拉曼光谱波长点的范围。
  9. 根据权利要求3所述的方法，其特征在于，所述步骤S8中基于偏最小二乘回归方法建立泡沫层拉曼光谱与钨浮选精矿品位间的回归模型，是基于偏最小二乘回归方法对钨浮选泡沫层样本的代表性拉曼光谱筛选特征谱段后的数据集合进行分析处理从而建立定量模型，包含以下三个步骤：
    第一步，随机选取总样本中的70％作为训练集，而剩余30％样本作为测试集，将其划分成训练样本集和测试样本集两个光谱数据集；
    第二步，基于训练样本集，采用偏最小二乘回归法建立钨精矿品位和拉曼光谱的对应回归模型；
    第三步，通过测试样本对已建立的定量模型进行验证：将测试样本的拉曼光谱数据代入定量模型中进行预测和拟合分析，通过预测值与实际值之间的误差来判断模型性能，若两者之间的误差超过实际浮选过程中精矿品位检测所要求的误差，则通过增大建模样本数、调整偏最小二乘模型中的主成分个数的方式调整模型参数，直至误差符合要求，从而优化回归模型。
  10. 根据权利要求9所述的方法，其特征在于，所述步骤S8中第二步，采用偏最小二乘回归法建立钨精矿品位和拉曼光谱的对应回归模型包括如下步骤：
    步骤(1)：首先对输入的训练集样本的拉曼光谱矩阵进行主成分分解，其后利用光谱主成分得分矩阵和精矿品位向量进行回归，其中对光谱矩阵X进行分解的公式如下：
    Figure PCTCN2020097212-appb-100002
    公式中，QT为光谱矩阵X分解得到的得分矩阵，而P则为X分解得到的载荷矩阵，q k和p k分别为矩阵QT和P的第k列列向量，l为上述矩阵的列数，而E X则为应用偏最小二乘法拟合矩阵X所产生的偏差，矩阵符号T为求矩阵的转置；
    步骤(2)：基于光谱矩阵分解后所得到的得分矩阵QT及样本的精矿品位向量Y进行回归，从而得到模型的关联矩阵B计算如下：
    B＝(Q TQ) -1Q TY
    步骤(3)：利用定量分析模型对测试集样本的精矿品位进行预测，具体步骤为根据测试集样本的光谱矩阵X′以及模型的载荷矩阵P求出其得分矩阵Q′，从而最终得到其精矿品位预测值如下：
    y＝(X′) TB′
    其中，B′＝[(Q′) T(Q′)] -1(Q′) TY。
PCT/CN2020/097212 2020-04-02 2020-06-19 用于钨浮选精矿品位在线检测的光纤拉曼系统及方法 WO2021196416A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010255653.X 2020-04-02
CN202010255653.XA CN111398243A (zh) 2020-04-02 2020-04-02 用于钨浮选精矿品位在线检测的光纤拉曼系统及方法

Publications (1)

Publication Number Publication Date
WO2021196416A1 true WO2021196416A1 (zh) 2021-10-07

Family

ID=71433579

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/097212 WO2021196416A1 (zh) 2020-04-02 2020-06-19 用于钨浮选精矿品位在线检测的光纤拉曼系统及方法

Country Status (2)

Country Link
CN (1) CN111398243A (zh)
WO (1) WO2021196416A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114113038A (zh) * 2021-12-01 2022-03-01 西安近代化学研究所 一种通过拉曼光谱快速测定硝硫混酸组分含量的方法
CN114354571A (zh) * 2021-12-22 2022-04-15 中国科学院合肥物质科学研究院 一种基于半峰宽与峰高的易制毒化学品拉曼特征峰识别方法
CN114383985A (zh) * 2021-12-24 2022-04-22 鞍钢集团矿业有限公司 基于多参数特征融合的浮选精矿品位检测方法
CN114707540A (zh) * 2022-03-14 2022-07-05 青岛明华电子仪器有限公司 一种谱图基线的确定方法
CN114813434A (zh) * 2022-02-16 2022-07-29 浙江工业大学 基于多目标优化和多光谱技术的水分和粒径同时检测的方法
CN114912348A (zh) * 2022-04-15 2022-08-16 东莞理工学院 一种光纤sers探针中光纤拉曼背景去除的方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112161959B (zh) * 2020-09-25 2022-10-25 山东达峰海洋科技有限公司 一种基于荧光光谱全局重构的叶绿素浓度测量方法
CN112730373A (zh) * 2020-12-03 2021-04-30 北京信息科技大学 一种用于深度学习训练的拉曼光谱数据集分析方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102495042A (zh) * 2011-12-09 2012-06-13 湖南大学 一种粉末混合物的拉曼光谱准确定量分析方法
CN104655607A (zh) * 2015-01-29 2015-05-27 中南大学 一种浮选泡沫层硫化锑矿物品位检测方法
CN105300954A (zh) * 2015-09-02 2016-02-03 中南大学 一种锑矿浮选非均质泡沫层矿物的拉曼光谱表征方法
US20160103070A1 (en) * 2014-10-08 2016-04-14 Samsung Electronics Co., Ltd. Method for multiple analysis of raman spectroscopy signal
CN108844941A (zh) * 2018-05-30 2018-11-20 武汉工程大学 一种基于拉曼光谱和pca-hca的不同品位磷矿的鉴别和分类方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1140792C (zh) * 2001-05-16 2004-03-03 昆明理工大学 一种宝石矿物的拉曼识别法
DE102005031140A1 (de) * 2005-07-04 2007-01-18 Erwin Kayser-Threde Gmbh Verfahren und Vorrichtung zur stofflichen Identifizierung von Oberflächen
JP5225235B2 (ja) * 2009-08-24 2013-07-03 株式会社エージーティ・ジェム・ラボラトリー 宝石物質の鑑別方法
DE112013004743T5 (de) * 2012-09-26 2015-09-24 Panalytical Inc. Multi-Sensoranalyse eines komplexen geologischen Materials

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102495042A (zh) * 2011-12-09 2012-06-13 湖南大学 一种粉末混合物的拉曼光谱准确定量分析方法
US20160103070A1 (en) * 2014-10-08 2016-04-14 Samsung Electronics Co., Ltd. Method for multiple analysis of raman spectroscopy signal
CN104655607A (zh) * 2015-01-29 2015-05-27 中南大学 一种浮选泡沫层硫化锑矿物品位检测方法
CN105300954A (zh) * 2015-09-02 2016-02-03 中南大学 一种锑矿浮选非均质泡沫层矿物的拉曼光谱表征方法
CN108844941A (zh) * 2018-05-30 2018-11-20 武汉工程大学 一种基于拉曼光谱和pca-hca的不同品位磷矿的鉴别和分类方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LI YUAN-YING;XU DE-GANG;GUI WEI-HUA;YANG CHUN-HUA;CAI YAO-YI: "A Novel Method for Measuring Antimony Sulfide Content Based on Gaussian-Peak Fitting of Raman Spectroscopy", SPECTROSCOPY AND SPECTRAL ANALYSIS, vol. 37, no. 12, 15 December 2017 (2017-12-15), CN, pages 3743 - 3748, XP055855149, ISSN: 1000-0593, DOI: 10.3964/j.issn.1000-0593(2017)12-3743-06 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114113038A (zh) * 2021-12-01 2022-03-01 西安近代化学研究所 一种通过拉曼光谱快速测定硝硫混酸组分含量的方法
CN114354571A (zh) * 2021-12-22 2022-04-15 中国科学院合肥物质科学研究院 一种基于半峰宽与峰高的易制毒化学品拉曼特征峰识别方法
CN114354571B (zh) * 2021-12-22 2024-05-28 中国科学院合肥物质科学研究院 一种基于半峰宽与峰高的易制毒化学品拉曼特征峰识别方法
CN114383985A (zh) * 2021-12-24 2022-04-22 鞍钢集团矿业有限公司 基于多参数特征融合的浮选精矿品位检测方法
CN114383985B (zh) * 2021-12-24 2023-09-15 鞍钢集团矿业有限公司 基于多参数特征融合的浮选精矿品位检测方法
CN114813434A (zh) * 2022-02-16 2022-07-29 浙江工业大学 基于多目标优化和多光谱技术的水分和粒径同时检测的方法
CN114707540A (zh) * 2022-03-14 2022-07-05 青岛明华电子仪器有限公司 一种谱图基线的确定方法
CN114707540B (zh) * 2022-03-14 2024-05-31 青岛明华电子仪器有限公司 一种谱图基线的确定方法
CN114912348A (zh) * 2022-04-15 2022-08-16 东莞理工学院 一种光纤sers探针中光纤拉曼背景去除的方法

Also Published As

Publication number Publication date
CN111398243A (zh) 2020-07-10

Similar Documents

Publication Publication Date Title
WO2021196416A1 (zh) 用于钨浮选精矿品位在线检测的光纤拉曼系统及方法
CN105092540B (zh) 一种食用油中重金属铅含量的快速高精度检测方法
CN102053083B (zh) 一种基于偏最小二乘法的煤质特性在线测量方法
CN104062257B (zh) 一种基于近红外光谱测定溶液中总黄酮含量的方法
CN101509872B (zh) 一种基于回归分析的煤质在线检测分析方法
CN109085136B (zh) 近红外漫反射光谱测定水泥生料氧化物成分含量的方法
CN104062256B (zh) 一种基于近红外光谱的软测量方法
CN102313731B (zh) 一种未知物组成元素含量在线检测方法
US20060291619A1 (en) Method and Apparatus for Material Identification
CN102410993B (zh) 基于激光诱导等离子体发射光谱标准化的元素测量方法
CN102004097B (zh) 一种基于主导因素结合偏最小二乘法的煤质在线检测方法
CN104655607B (zh) 一种浮选泡沫层硫化锑矿物品位检测方法
CN105092436B (zh) 一种沉积物粒度光谱分析方法及装置
CN113155809A (zh) 一种矿石分类与实时定量分析的光谱检测新方法
CN105784672A (zh) 一种基于双树复小波算法的毒品检测仪标准化方法
CN102410992B (zh) 简化的激光诱导等离子体光谱标准化的元素测量方法
CN107247033A (zh) 基于快速衰减式淘汰算法和plsda鉴别黄花梨成熟度的方法
CN104596975A (zh) 近红外漫反射光谱技术测定造纸法再造烟叶木质素的方法
CN104596979A (zh) 近红外漫反射光谱技术测定造纸法再造烟叶纤维素的方法
Cai et al. Online detection of concentrate grade in the antimony flotation process based on in situ Raman spectroscopy combined with a CNN-GRU hybrid model
Legnaioli et al. Univariate linear methods
CN109001182A (zh) 封闭容器中酒精含量的拉曼光谱无损测定方法
CN103134764B (zh) 由透射红外光谱预测原油实沸点蒸馏曲线的方法
CN104596982A (zh) 近红外漫反射光谱技术测定造纸法再造烟叶果胶的方法
CN110579467B (zh) 一种时间分辨激光诱导击穿光谱定量方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20929054

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20929054

Country of ref document: EP

Kind code of ref document: A1