LIBS-based bauxite component quantitative detection method and related apparatus

Abstract

An LIBS-based bauxite component quantitative detection method and a related apparatus. The method comprises: controlling an LIBS system to collect actual spectral data of bauxite under test; preprocessing the actual spectral data to obtain preprocessed actual spectral data; on the basis of the preprocessed actual spectral data, extracting corresponding feature elements to be detected, said feature elements comprising one or more of information about feature wavelengths to be detected, information about element wavelengths to be detected, and information about remaining wavelengths to be detected; and, on the basis of said feature elements and a quantitative detection model, determining component information of said bauxite.

Description

LIBS-based bauxite composition quantitative detection method and related equipment

[0001] Cross-reference to related applications

[0002] This application claims priority to Chinese Patent Application No. 202410646415X, filed on May 23, 2024, the entire contents of which are incorporated herein by reference. TECHNICAL FIELD

[0003] The present disclosure relates to the field of ore detection, and in particular, to a LIBS-based bauxite composition quantitative detection method and related equipment. BACKGROUND

[0004] Bauxite is the main raw material for alumina production, with complex composition and large variation in chemical composition. The main components include Al2O3, SiO2, Fe2O3, TiO2, CaO, MgO, K2O, Na2O, etc., as well as Ga, Li, Ge, Nb, and other trace elements. The preparation of raw ore slurry is the first process in alumina production, which involves preparing and dispersing bauxite, lime, sodium aluminate solution, and recycled mother liquor into a raw ore slurry that meets the dissolution requirements. The composition of bauxite determines the optimal proportion of other materials, and fluctuations in composition will result in excess or insufficient addition of other materials, affecting the entire alumina production line, causing fluctuations in actual dissolution and decomposition rate, increased material and energy consumption, and decreased product quality.

[0005] Traditional LIBS-based (Laser-Induced Breakdown Spectroscopy) bauxite composition quantitative detection methods include chemical analysis and spectral analysis. These methods have limitations such as complex sample pretreatment, long analysis time, need for large amounts of chemical reagents, tedious operation, and susceptibility to human error. LIBS is a new analysis technology that forms a plasma on the surface of a sample by focusing an ultra-short pulse laser, and then analyzes the plasma emission spectrum to determine the material composition and content of the sample. The energy density after focusing the ultra-short pulse laser is high, which can excite any state of sample to form a plasma. In principle, LIBS technology can analyze any state of sample, limited only by the power of the laser and the sensitivity and wavelength range of the spectrometer and detector. It is commonly used to detect the elemental composition and relative abundance of substances, with characteristics such as fast detection speed, no need for complex sample pretreatment, and simultaneous detection of multiple elements. It is increasingly widely used in metallurgy, materials, environmental monitoring, rock and mineral, and industrial online analysis fields. Compared with traditional spectral analysis techniques, LIBS technology has simple sample handling and is more suitable for the determination of bauxite and other difficult-to-decompose samples in wet processes.

[0006] The technical field relates to a laser-induced breakdown spectroscopy (LIBS) based bauxite composition quantitative detection method and system, which can reduce sample processing steps, quickly analyze bauxite composition, improve detection efficiency, and further improve the effectiveness of automatic ore blending in alumina production.

[0007] SUMMARY

[0008] A series of simplified concepts are introduced in the summary section, which will be further described in detail in the detailed description. The summary section of the present disclosure does not mean to attempt to limit the key features and essential technical features of the claimed technical solutions, nor to determine the protection scope of the claimed technical solutions.

[0009] In a first aspect, the present disclosure provides a LIBS based bauxite composition quantitative detection method, comprising:

[0010] controlling a LIBS system to collect actual spectral data of a to-be-detected bauxite; performing a preprocessing operation on the actual spectral data to obtain preprocessed actual spectral data; extracting corresponding to-be-detected characteristic elements based on the preprocessed actual spectral data, wherein the to-be-detected characteristic elements include one or more of to-be-detected characteristic wavelength information, to-be-detected element wavelength information, and to-be-detected remaining wavelength information; and determining composition information of the to-be-detected bauxite according to the to-be-detected characteristic elements and a quantitative detection model, wherein the quantitative detection model is a machine learning model, and the quantitative detection model is used to perform feature matching according to the to-be-detected characteristic elements to determine the composition information of the to-be-detected bauxite.

[0011] In a second aspect, the present disclosure further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to execute the computer program stored in the memory to implement the steps of the LIBS based bauxite composition quantitative detection method.

[0012] In a third aspect, the present disclosure further provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the LIBS based bauxite composition quantitative detection method. BRIEF DESCRIPTION OF DRAWINGS

[0013] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The accompanying drawings are included to provide a description of preferred embodiments, and are not meant to limit the present disclosure. Moreover, the same reference numerals are used throughout the several drawings to represent similar or same components. In the drawings:

[0014] FIG. 1 is a flowchart of a method for quantitatively detecting the composition of bauxite based on LIBS according to some embodiments of the present disclosure;

[0015] FIG. 2 is a schematic diagram of a LIBS system according to some embodiments of the present disclosure;

[0016] FIG. 3 is a LIBS spectrum of bauxite according to some embodiments of the present disclosure;

[0017] FIG. 4 is a model structure diagram for modeling two types of wavelength data respectively according to some embodiments of the present disclosure;

[0018] FIG. 5 is a comparison diagram of the results of detecting partial elements according to some embodiments of the present disclosure;

[0019] FIG. 6 is a schematic diagram of an electronic device for quantitatively detecting the composition of bauxite based on LIBS according to some embodiments of the present disclosure. DETAILED DESCRIPTION

[0020] The terms "first", "second", "third", "fourth" and the like in the description and claims of the present disclosure and above-mentioned drawings, if any, are used for distinguishing between similar objects, and do not necessarily have a particular chronological, sequential, or hierarchical order to them. It is to be understood that the data used in this way can be interchanged, where appropriate, so that the embodiments described herein can be carried out in sequences other than those illustrated or otherwise described herein. Moreover, the terms "comprise" and "have" and any variations thereof are intended to cover a non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a list of steps or units can not necessarily be limited to those listed steps or units but can include other not-listed steps or units, or can inherently include other not-listed steps or units. The technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure, not all the embodiments.

[0021] Referring to FIG. 1, it is a flowchart of a method for quantitatively detecting the composition of bauxite based on LIBS according to some embodiments of the present disclosure. The method for quantitatively detecting the composition of bauxite based on LIBS according to some embodiments of the present disclosure can include steps S110-S140.

[0022] In step S110, the actual spectral data of the bauxite to be detected is collected by controlling the LIBS system.

[0023] In some embodiments, the LIBS system irradiates the bauxite sample surface with a laser, excites the sample to generate plasma, and acquires the elemental composition information of the sample by detecting the spectrum emitted by the plasma. This step involves controlling the LIBS system to laser excite the bauxite sample to be measured and collect its actual spectral data.

[0024] In step S120, the actual spectral data is preprocessed to obtain preprocessed actual spectral data.

[0025] In some embodiments, the preprocessing operation can include but is not limited to signal denoising, background subtraction, spectral normalization, etc. The purpose of the preprocessing operation is to remove or reduce noise and interference in the actual spectrum, improve data quality, and improve analysis accuracy. In this step, the collected actual spectral data is preprocessed to obtain clean preprocessed actual spectral data.

[0026] In step S130, the corresponding to-be-measured characteristic elements are extracted based on the preprocessed actual spectral data, wherein the to-be-measured characteristic elements include one or more of to-be-measured characteristic wavelength information, to-be-measured element wavelength information, and to-be-measured remaining wavelength information.

[0027] In some embodiments, the characteristic information related to the target element is extracted from the preprocessed spectral data. The to-be-measured characteristic elements include to-be-measured characteristic wavelength information, to-be-measured element wavelength information, and to-be-measured remaining wavelength information. The to-be-measured characteristic wavelength information is the wavelength selected by the target element based on a specific characteristic selection method. The specific characteristic selection method includes one or more of principal component analysis load method, partial least squares regression coefficient method, and Pearson correlation analysis method.

[0028] The to-be-measured element wavelength information is the sensitive wavelength of the target element in a specific wavelength range. The to-be-measured element wavelength information refers to the wavelength range related to the target element in spectral analysis, which has higher sensitivity to detection. For each target element (such as Al, Si, Fe, etc.), its spectral characteristics are more obvious or have higher sensitivity in a certain wavelength region, i.e., the signal change at these wavelengths can better reflect the existence and concentration of the element. These specific wavelengths are called sensitive wavelengths. Selecting these sensitive wavelengths can improve the detection accuracy and signal specificity of the target element, so that the analysis process focuses more accurately on the characteristic signal of the target element, reducing the noise or interference that other wavelengths may introduce.

[0029] The to-be-measured remaining wavelength information is the wavelength information other than the to-be-measured element wavelength information. The target element is the element corresponding to the element component that needs to be measured, and the target element can include one or more of Al, Si, Fe, Ca, K, Na, Mg, and Ti.

[0030] In step S140, the composition information of the bauxite to be detected is determined according to the above-mentioned characteristic elements to be detected and the quantitative detection model, wherein the quantitative detection model is used to perform feature matching according to the characteristic elements to be detected to determine the composition information of the bauxite to be detected.

[0031] In some embodiments, the quantitative detection model is constructed based on a machine learning algorithm, mainly by training a large amount of known sample data to learn the feature pattern of each component. The model training stage usually includes a large amount of data of different bauxite samples. When determining the composition information of the bauxite to be detected, the machine learning model matches the detected characteristic elements, and this matching is performed by the feature pattern in the model. Finally, the model outputs the composition information of the sample to be detected according to the matching result of the features.

[0032] In summary, the present disclosure proposes a LIBS-based bauxite composition quantitative detection method based on laser-induced breakdown spectroscopy. Based on the laser-induced breakdown spectroscopy technology, the spectral data of the bauxite standard sample is collected, the actual spectral data is preprocessed, the characteristic wavelength information to be detected, the element wavelength information to be detected, and the remaining wavelength information to be detected are analyzed and screened, the combination wavelength data of one or more of the characteristic wavelength information to be detected, the element wavelength information to be detected, and the remaining wavelength information to be detected is used to construct an element quantitative detection modeling data set, a detection model is established by combining mechanism driving and data driving and using a machine learning method, and bauxite composition quantitative detection is realized. The bauxite composition quantitative detection system designed according to this method can save the sample processing step, quickly analyze the composition of bauxite, improve the detection efficiency, and further improve the effectiveness of automatic ore blending in alumina production.

[0033] In some embodiments, the quantitative detection model is obtained by the following steps:

[0034] The LIBS system is controlled to collect standard spectral data of a standard bauxite, and the composition of the standard bauxite is obtained.

[0035] The standard spectral data of the standard bauxite is preprocessed to obtain preprocessed standard spectral data of the standard bauxite.

[0036] extracting standard characteristic elements corresponding to the standard bauxite based on the pre-processed standard spectral data, the standard characteristic elements including one or more of standard characteristic wavelength information, standard element wavelength information, and standard residual wavelength information, the to-be-measured characteristic wavelength information and the standard characteristic wavelength information being wavelengths selected based on a specific characteristic selection method, the to-be-measured element wavelength information and the standard element wavelength information being sensitive wavelengths of a target element in a specific wavelength range, the to-be-measured residual wavelength information being wavelength information other than the to-be-measured element wavelength information, and the standard residual wavelength information being wavelength information other than the standard element wavelength information.

[0037] training the quantitative detection model based on the standard characteristic elements corresponding to the standard bauxite and bauxite composition of the standard bauxite.

[0038] In some embodiments, the quantitative detection model is established by analyzing standard spectral data of standard samples and extracting standard characteristic elements. The standard characteristic elements include standard characteristic wavelength information, standard element wavelength information, and standard residual wavelength information. The information of these standard characteristic elements is compared and analyzed with corresponding characteristic information of to-be-measured samples, and finally the specific composition of the to-be-measured bauxite is determined.

[0039] Specifically, the standard bauxite includes one or more, and the standard spectral data of the to-be-measured bauxite is collected by using a LIBS system. The collected spectral data is pre-processed, including but not limited to removing noise, baseline correction, spectral normalization, and the like, to enhance the quality of the data, so as to obtain more accurate pre-processed standard spectral data. Standard characteristic elements related to bauxite composition are extracted from the pre-processed spectral data. These standard characteristic elements include the following wavelength information, which corresponds to different spectral characteristics and is used to reflect the sensitivity and characteristics of the composition:

[0040] The standard characteristic wavelength information is an important wavelength related to a target element (such as aluminum, silicon, titanium, etc.) selected by a specific characteristic selection method, and is used to reflect the key spectral characteristics of the element.

[0041] The standard element wavelength information is a sensitive wavelength range of a specific element in the spectrum, which is usually a wavelength region where the spectral peak of the element is located, and reflects the strong response of the element in the spectrum.

[0042] The standard residual wavelength information is other wavelength information other than the standard element wavelength information, which is used to provide background information and other possible characteristics, so that the model can more comprehensively understand the spectral data.

[0043] The standard bauxite corresponding to the standard characteristic elements and the bauxite composition information of the standard bauxite are taken as a training set to perform model training. Finally, a quantitative detection model is obtained, which can identify the characteristic element wavelength information in the spectral data and correspond to the actual content of the bauxite composition.

[0044] The spectral data of the bauxite is collected by using the LIBS system, the characteristic wavelength information related to the composition content in the spectral data is extracted through preprocessing and feature extraction, and then an initial model is trained based on the characteristic data, and finally a quantitative detection model for accurately quantifying the composition of the bauxite is obtained.

[0045] It should be noted that the initial model corresponding to the quantitative detection model includes but is not limited to a machine learning model of single output or multiple outputs, partial least squares regression, multilayer perception, convolutional neural network and recurrent neural network.

[0046] In some embodiments, the specific feature selection method described above includes one or more of principal component analysis loading method, partial least squares regression coefficient method and Pearson correlation analysis method.

[0047] In some embodiments, the principal component analysis loading method (PCA-loading), the partial least squares regression coefficient method (PLS) or the Pearson correlation analysis method can be used when performing feature selection, and the element wavelength with large contribution to element composition detection is selected as the element characteristic wavelength.

[0048] The PCA-loading method can reflect the correlation between the principal component and the spectral wavelength. The greater the absolute value of the loading corresponding to the wavelength, the greater the contribution rate of the wavelength to the principal component.

[0049] The regression coefficient in the PLS method reflects the influence degree of the spectral wavelength on the composition information prediction. The greater the absolute value of the regression coefficient, the higher the relevance of the regression coefficient to the composition information.

[0050] The Pearson correlation analysis method calculates the correlation coefficient between the composition detection value and the spectral wavelength. The greater the absolute value of the correlation coefficient, the higher the correlation between the spectral wavelength and the detection value.

[0051] In some embodiments, the quantitative detection model described above includes a machine learning model of single output or multiple outputs, partial least squares regression, multilayer perception, convolutional neural network and recurrent neural network.

[0052] In some embodiments, partial least squares regression is a statistical method used to establish a model between one or more predictor variables (e.g., spectral wavelengths) and one or more response variables (e.g., elemental content). In LIBS analysis, single-output PLS can be used to predict a single component, while multi-output PLS can simultaneously predict multiple components. PLS improves the accuracy of prediction by analyzing the linear relationship between wavelengths and component content to extract the most informative features.

[0053] A multi-layer perceptron is a basic artificial neural network that consists of an input layer, one or more hidden layers, and an output layer. MLPs are capable of capturing and modeling non-linear relationships, which is particularly useful when dealing with complex spectral data. In LIBS analysis, MLPs can accurately predict the composition of unknown samples by learning the complex patterns between input spectral data and known component content.

[0054] Convolutional neural networks (CNNs) are primarily used for image processing but are also applicable to any form of spatial data analysis. In LIBS spectral analysis, CNNs can effectively process and analyze the local patterns and features of spectral data. By utilizing convolutional layers to capture local dependencies between wavelengths, CNNs are able to extract features that are crucial for component identification and quantification.

[0055] Recurrent neural networks are particularly well-suited for handling sequential data, such as time-series data or spectral data. In LIBS analysis, RNNs (Recurrent Neural Networks) can effectively identify patterns and trends in spectral data by processing the sequential dependencies of wavelength data. This makes RNNs excel in predicting chemically components that change continuously or dynamically.

[0056] Machine learning models are capable of handling the vast amount of data generated by LIBS systems and learning from it how to predict the chemical composition of different samples. By applying these advanced models, the analysis process not only becomes more automated but also more accurate and reliable. The efficiency and flexibility of machine learning models in handling spectral data make them indispensable tools in modern materials analysis, especially when dealing with large-scale datasets to uncover complex chemical and physical properties

[0057] In some embodiments, the quantitative detection model described above is evaluated and tuned based on the root mean square error.

[0058] In some embodiments, the model training uses the root mean square error (RMSE) for model evaluation and preferably model parameter tuning.

[0059] The root mean square error (RMSE) calculation method is as follows:

[0060] Where y is the known composition information, and y' is the composition information detected by the model.

[0061] The preferred model parameters refer to the hyperparameters of the model optimized using the grid search method.

[0062] In some embodiments, the specific wavelength range is 200-837 nm.

[0063] In some embodiments, the wavelength range of 200-837 nm covers the spectral region commonly used in LIBS technology, and these wavelengths can be used to detect a variety of different elements. Through this range, the characteristic emission lines of a variety of elements can be captured for subsequent analysis and identification.

[0064] In some embodiments, the preprocessing operation includes one or more of asymmetric partial least squares correction, wavelet transform, derivative method, and polynomial transform.

[0065] In some embodiments, the preprocessing operation refers to the preliminary processing of the collected spectral data to improve data quality and increase the accuracy and reliability of the analysis. The preprocessing operation can include but is not limited to partial least squares correction, wavelet transform, derivative method, and polynomial transform.

[0066] The asymmetric partial least squares correction is used to correct the baseline drift in the spectral data and improve the data quality. The wavelet transform effectively removes noise and extracts useful signals by performing wavelet transform on the data. The derivative method calculates the derivative of the spectral data to enhance the spectral features and facilitate the identification of characteristic wavelengths. The polynomial transform uses polynomial transform to smooth the data and correct the baseline.

[0067] In some embodiments, the quantitative detection model is established based on the characteristics of the standard characteristic elements. When the standard characteristic elements include multiple information, a model is established based on the combination of multiple information or a model is established based on a single information and then weighted modeling is performed.

[0068] In some embodiments, when the standard characteristic elements include multiple information, the quantitative detection model can use the convolutional neural network model as described in FIG. 4, which includes two groups of three convolutional structures. The features of the last convolutional structure in each group are flattened and then combined. The quantitative detection model also includes a feature weighting structure for feature weighting. Each convolutional structure consists of a one-dimensional convolutional layer, an activation layer, a batch normalization layer, and a one-dimensional max pooling layer. The feature weighting structure includes three fully connected layers and two activation layers.

[0069] In some embodiments, the model is constructed using a convolutional neural network, which specifically includes:

[0070] Each group of convolutional structures contains several convolutional layers, each followed by an activation layer, a batch normalization layer, and a one-dimensional max-pooling layer. The convolutional layers are used to extract local features in the spectral data, the batch normalization layers are used to speed up the training process, and the max-pooling layers are used to reduce the feature dimension. After the last convolutional structure in each group, the features are flattened and merged, which helps to fuse information from different convolutional structures to provide a more comprehensive data representation. The feature weighting structure includes 3 fully connected layers and 2 activation layers, which are used to further process and weight the merged features to optimize the final output and improve the prediction accuracy of the model.

[0071] In some embodiments, the target elements include one or more of Al, Si, Fe, Ca, K, Na, Mg, and Ti.

[0072] In some embodiments, as shown in FIG. 2, a laser-induced breakdown spectroscopy-based bauxite composition quantitative detection system that can be used in the present disclosure includes a laser, a spectrometer, a detector, a delay timer, and a processing terminal. The system is used to collect bauxite spectral data. The hardware modules are shown in FIG. 2 and include a laser, a spectrometer, an IsCMOS camera, a timing controller, and an industrial computer. The wavelength of the laser is 1064 nm, the single pulse energy is 100 mJ, and the repetition frequency is 1 Hz; the spectrometer is a four-channel spectrometer with a wavelength range of 200 nm to 837 nm and a resolution of 0.1 nm; the IsCMOS camera has 2048 pixels; the timing controller sets the acquisition delay to 300 ns; and the industrial computer has strong noise resistance and stability.

[0073] The specific steps of the laser-induced breakdown spectroscopy-based bauxite composition quantitative detection include steps 201 to 209:

[0074] Step 201: Collect spectral data of 60 bauxite standard samples with known compositions using a LIBS system.

[0075] The LIBS system includes a laser, a spectrometer, an IsCMOS camera, a timing controller, and an industrial computer, as shown in FIG. 2. The Al, Si, Fe, Ca, K, Na, Mg, and Ti element contents of the 60 bauxite standard samples are known. When collecting spectral data, the LIBS system is used to laser excite the surface of each sample, and each sample is excited 100 times, and the 100 spectral data are averaged. The spectral range of the spectral data is 200 nm to 837 nm, with a total of 8192 spectral wavelengths, and the data size is (8192, 60).

[0076] Step 202: Preprocessing the spectral data to eliminate matrix offset and noise interference. Asymmetric partial least squares correction method, wavelet transform method, derivative method or polynomial transformation method are used to eliminate matrix offset and noise interference. In this embodiment, the asymmetric partial least squares correction method is used to preprocess the spectral data to eliminate matrix offset and noise interference. The asymmetric partial least squares correction method has an asymmetric factor of 0.001, a threshold of 0.05, a smoothing factor of 4, and an iteration number of 10. The pretreated bauxite spectral data is shown in FIG. 3.

[0077] Step 203: Extracting the element wavelengths of Al, Si, Fe, Ca, K, Na, Mg and Ti elements. The element wavelength refers to all sensitive wavelengths of Al, Si, Fe, Ca, K, Na, Mg and Ti elements between 200 nm and 837 nm. Among them, 27 Al element wavelengths, 18 Si element wavelengths, 132 Fe element wavelengths, 32 Ca element wavelengths, 6 K element wavelengths, 14 Na element wavelengths, 23 Mg element wavelengths and 95 Ti element wavelengths are extracted.

[0078] Step 204: Combining the known composition information of the bauxite sample, and selecting the characteristic wavelengths of Al, Si, Fe, Ca, K, Na, Mg and Ti elements by using the characteristic selection method. When performing characteristic selection, the PCA-loading method (PCA-loading), the PLS method (PLS) or the Pearson correlation analysis method can be used to select the element wavelengths with large contribution to element composition detection as element characteristic wavelengths.

[0079] In this embodiment, the PCA-loading method is used for characteristic selection, and the loadings of the three principal components with the highest contribution rate in PCA analysis are selected. The characteristic spectra of Al, Si, Fe, Ca, K, Na, Mg and Ti elements obtained by selection are shown in Table 1.

[0080] Table 1

[0081] Step 205: Combining the characteristic wavelengths, element wavelengths and remaining wavelengths to construct the modeling data set of Al, Si, Fe, Ca, K, Na, Mg and Ti elements. In this application, the wavelengths are divided into three categories, namely characteristic wavelengths, element wavelengths and remaining wavelengths. The characteristic wavelength refers to the wavelength selected from the element wavelengths of different elements in step 4; the element wavelength refers to the sensitive wavelength of different elements extracted in step 3; and the remaining wavelength refers to the wavelength remaining after the element wavelength is extracted in step 3.

[0082] The modeling data set is constructed by one of the following four methods:

[0083] (1) Only using characteristic wavelengths; for example, the data set size of Al element is (8, 60); for example, the data set size of Si element is (7, 60); for example, the data set size of Fe element is (24, 60);

[0084] (2) Only using element wavelengths; for example, the data set size of Al element is (27, 60); for example, the data set size of Si element is (18, 60); for example, the data set size of Fe element is (132, 60);

[0085] (3) Combination of characteristic wavelengths and residual wavelengths; for example, the data set size of Al element is (8+8165, 60); for example, the data set size of Si element is (7+8174, 60); for example, the data set size of Fe element is (24+8060, 60);

[0086] (4) Combination of element wavelengths and residual wavelengths; for example, the data set size of Al element is (27+8165, 60); for example, the data set size of Si element is (18+8174, 60); for example, the data set size of Fe element is (132+8060, 60).

[0087] The modeling data set is divided into a training set and a validation set in a ratio of 7:3.

[0088] Step 206: Constructing detection models of Al, Si, Fe, Ca, K, Na, Mg and Ti elements. The detection model refers to a machine learning model capable of realizing regression prediction, such as a single output or multiple outputs of partial least squares regression, multilayer perception, convolutional neural network or recurrent neural network, etc. In the process of constructing the detection model, the modeling data set used is one of the four data sets established in step 205; if the modeling data set is a multi-class wavelength combination data set, two types of wavelength data are combined for modeling or two types of wavelength data are modeled respectively and then the models are weighted.

[0089] In this embodiment, the modeling data set is constructed by combining element wavelengths and residual wavelengths, and two types of wavelength data can be combined for modeling or two types of wavelength data can be modeled respectively and then the models can be weighted, and in this embodiment, two types of wavelength data are combined for modeling, the detection model constructed is a partial least squares regression model (PLSR), and eight elements are respectively used to construct detection models by using corresponding modeling data sets.

[0090] PLSR combines the characteristics of principal component analysis, canonical correlation analysis and linear regression analysis in the modeling process, provides a method of multiple-to-multiple linear regression modeling, and has good performance especially when the number of two groups of variables is large and there is multiple correlation, and the sample size is small.

[0091] In addition, a feasible method of establishing a detection model by weighting the two types of wavelength data after modeling is listed in this embodiment. Two types of wavelength are used to establish convolutional neural network models for feature extraction. The features of the last layer are selected for feature merging and weighted training to obtain an element composition detection model. The model structure is shown in FIG. 4. The element wavelength data set and the residual wavelength data set are respectively subjected to three-layer convolutional structure for feature extraction. The features of the last convolutional structure are flattened and then subjected to feature merging. Three fully connected layers are used for feature weighting, and finally the element detection composition is output. The convolutional structure is composed of a one-dimensional convolutional layer, an activation layer, a batch normalization layer and a one-dimensional maximum pooling layer. The feature weighting structure is composed of three fully connected layers and two activation layers.

[0092] Step 207: model training. The root mean square error (RMSE) is used for model evaluation and preferred model parameters. The root mean square error (RMSE) calculation method is as follows:

[0093] Where y is the known composition information, and y' is the composition information detected by the model.

[0094] In this embodiment, PLSR is used to establish a detection model. The parameters for model training and optimization include the maximum number of iterations, the convergence standard and the number of principal components of PLSR. The maximum number of iterations is set to 1e4, and the convergence standard is set to 1e-3. FIG. 5 is a comparison chart of the validation set results of the elements Al, Fe and Si with the highest content in bauxite after the model training is completed. The validation set RMSE of Al element is 0.6990, the validation set RMSE of Si element is 0.1388, and the validation set RMSE of Fe element is 0.6157.

[0095] Step 208: Collect the spectral data of bauxite on the conveying belt before the alumina enters the mill. When collecting the spectral data of bauxite on the conveying belt, the sample is excited once, the conveying belt runs at a speed of 2 m / s, and one spectral data is collected every 2 m.

[0096] Step 209: Perform spectral data preprocessing according to step 202, and construct a detection data set according to step 205. According to 8 elements, 8 detection data sets are generated. The detection data sets of Al, Si, Fe, Ca, K, Na, Mg and Ti elements have 103, 95, 192, 107, 88, 92, 100 and 158 wavebands, respectively.

[0097] Step 210: model operation, to obtain the bauxite composition quantitative detection result. After one spectrum data is operated by the model, the quantitative detection values of Al, Si, Fe, Ca, K, Na, Mg and Ti elements are obtained.

[0098] As shown in FIG. 6, the embodiment of the present disclosure further provides an electronic device 300, which comprises a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and capable of running on the processor. When the processor 320 executes the computer program 311, the steps of any method for the LIBS-based bauxite composition quantitative detection described above are implemented.

[0099] In summary, the present disclosure provides a LIBS-based bauxite composition quantitative detection method and system based on laser-induced breakdown spectroscopy. Based on the laser-induced breakdown spectroscopy technology, the spectrum data of the bauxite standard sample is collected, the spectrum data is preprocessed, the element wavelength, characteristic wavelength and residual wavelength are analyzed and selected, the wavelength data is combined, the element quantitative detection modeling data set is constructed, the detection model is established by using the machine learning method in combination with the mechanism driving and data driving, and the bauxite composition quantitative detection is realized. Meanwhile, the bauxite composition quantitative detection system is designed, which can save the sample processing step, quickly analyze the bauxite composition, improve the detection efficiency, and further improve the effectiveness of the automatic ore blending of the alumina production.

[0100] Since the electronic device described in the embodiment is the device used to implement the LIBS-based bauxite composition quantitative detection device in the embodiment of the present disclosure, the specific implementation of the electronic device and its various forms can be understood by those skilled in the art based on the method described in the embodiment of the present disclosure. Therefore, how the electronic device implements the method in the embodiment of the present disclosure will not be described in detail here, as long as the device used by those skilled in the art to implement the method in the embodiment of the present disclosure belongs to the scope of the present disclosure.

[0101] In the specific implementation process, the computer program 311 can implement any embodiment of the corresponding embodiment of FIG. 1 when executed by the processor.

[0102] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0103] Those skilled in the art will appreciate that embodiments of the disclosure can be provided as methods, systems, or computer program products. Accordingly, the disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the disclosure can take the form of a computer program product on one or more computer readable storage media (including, but not limited to, disk memory, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0104] The disclosure is described in reference to the flowchart and / or block diagrams of the method, apparatus (system) and computer program product according to embodiments of the disclosure. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as a combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus generate means for implementing the functions specified in one or more flows in the flowchart and / or one or more blocks in the block diagram.

[0105] These computer program instructions can also be stored in a computer readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction means, which implements the functions specified in one or more flows in the flowchart and / or one or more blocks in the block diagram.

[0106] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide steps for implementing the functions specified in one or more flows in the flowchart and / or one or more blocks in the block diagram.

[0107] Embodiments of the disclosure also provide a computer program product, which includes computer software instructions, when the computer software instructions are run on a processing device, cause the processing device to perform the flow of LIBS-based quantitative detection of bauxite composition in the corresponding embodiments.

[0108] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions according to the embodiments of the present disclosure are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that the computer can store or be integrated into a data storage device such as a server, data center, etc. containing one or more available media sets. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)), etc.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0110] In several embodiments provided by the present disclosure, it should be understood that the disclosed system, device and method can be implemented by other manners. For example, the above-described device embodiments are only schematic, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the shown or discussed units can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0111] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0112] In addition, each function unit in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software function unit.

[0113] When the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present disclosure, essentially or in part, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present disclosure. The foregoing storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and various other media that can store program codes.

[0114] The above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than limit them; although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present disclosure.

Claims

1. A method for quantitatively detecting the composition of bauxite based on LIBS, comprising: controlling a LIBS system to collect actual spectral data of a bauxite to be detected; performing a pretreatment operation on the actual spectral data to obtain pretreated actual spectral data; extracting corresponding characteristic elements to be detected based on the pretreated actual spectral data, wherein the characteristic elements to be detected include one or more of characteristic wavelength information to be detected, element wavelength information to be detected, and remaining wavelength information to be detected; determining the composition information of the bauxite to be detected according to the characteristic elements to be detected and a quantitative detection model, wherein the quantitative detection model is used to perform feature matching according to the characteristic elements to be detected to determine the composition information of the bauxite to be detected. 2.The method for quantitatively detecting the composition of bauxite based on LIBS according to claim 1, wherein the quantitative detection model is obtained by the following steps: controlling a LIBS system to collect standard spectral data of a standard bauxite, and obtaining the composition of the standard bauxite; performing a pretreatment operation on the standard spectral data of the standard bauxite to obtain pretreated standard spectral data of the standard bauxite; extracting corresponding standard characteristic elements of the standard bauxite based on the pretreated standard spectral data, wherein the standard characteristic elements include one or more of standard characteristic wavelength information, standard element wavelength information, and standard remaining wavelength information, the characteristic wavelength information to be detected and the standard characteristic wavelength information are wavelengths selected based on a specific feature selection method, the element wavelength information to be detected and the standard element wavelength information are sensitive wavelengths of a target element in a specific wavelength range, the remaining wavelength information to be detected is wavelength information other than the element wavelength information to be detected, and the standard remaining wavelength information is wavelength information other than the standard element wavelength information; training the quantitative detection model based on the standard characteristic elements of the standard bauxite and the composition of the standard bauxite.

3. The LIBS-based quantitative detection method of bauxite composition according to claim 2, wherein, The specific feature selection method includes one or more of principal component analysis load method, partial least squares regression coefficient method, and Pearson correlation analysis method.

4. The LIBS-based quantitative detection method of bauxite composition according to claim 1, wherein, The quantitative detection model includes a machine learning model of single output or multiple outputs, partial least squares regression, multilayer perception, convolutional neural network, and recurrent neural network.

5. The LIBS-based quantitative detection method of bauxite composition according to claim 1, wherein, The quantitative detection model is evaluated and optimized based on root mean square error.

6. The LIBS-based quantitative detection method of bauxite composition according to claim 2, wherein, The specific wavelength range is 200nm-837nm.

7. The LIBS-based quantitative detection method of bauxite composition according to claim 1, wherein, The pretreatment operation includes one or more of asymmetric partial least squares correction method, wavelet transform method, derivative method, and polynomial transform method.

8. The LIBS-based quantitative detection method of bauxite composition according to claim 1, characterized in that, The quantitative detection model is established based on standard characteristic elements, and when the standard characteristic elements include multiple types of information, a model is built based on the combination of multiple types of information or a model is built based on a single type of information and then weighted modeling is performed.

9. The LIBS-based quantitative detection method of bauxite composition according to claim 1, wherein, The target element includes one or more of Al, Si, Fe, Ca, K, Na, Mg, and Ti.

10. An electronic device comprising: A memory and a processor, wherein the processor is configured to implement the steps of the method for quantitative detection of bauxite composition based on LIBS according to any one of claims 1-9 when executing a computer program stored in the memory.

11. A computer readable storage medium having stored thereon a computer program, wherein, The computer program is configured to implement the steps of the method for quantitative detection of bauxite composition based on LIBS according to any one of claims 1-9 when executed by the processor.

Images