A mine sample analysis method and system based on spectral correction

Abstract

The application provides a mine sample analysis method and system based on spectrum correction, and relates to the technical field of mine composition analysis. First, a first mine sample is pretreated, and XRF energy spectrum and gamma energy spectrum are obtained respectively, then fusion processing is performed on the two energy spectrums to obtain a first spectrum fusion feature vector, mineral combination analysis information of the first mine sample is obtained based on machine learning model analysis, finally, at least one mine sample with a similarity meeting a preset requirement is found in historical mine sample data according to the mineral combination analysis information, and collection parameters of a determined target mine sample are used as an optimal photon energy-time parameter set, analysis processing of the first mine sample is completed, and an analysis report is formed. The technical scheme of the application determines the best analysis parameters through two steps of artificial intelligence and big data screening, so that the mine analysis result is more accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mine composition analysis, and in particular to a mine sample analysis method and system based on spectral correction. BACKGROUND

[0002] With the increasing demand for detection accuracy and efficiency in the development of mine resources, traditional sample analysis technology is facing many challenges. Although the current mainstream X-ray fluorescence spectroscopy and photon activation analysis technology can detect light and heavy elements and precious metals respectively, there are problems such as large matrix effect interference, incomplete element coverage, and low efficiency of artificial correction. For example, XRF is prone to gold and silver element signal suppression when detecting high-lead content samples, and the high-energy gamma photon source of PAA needs to rely on complex equipment and has a long measurement period. At the same time, traditional spectral correction methods mostly use linear regression or physical models, which are difficult to adapt to the dynamic matrix characteristics such as complex mineral paragenetic relationship and uneven particle size distribution in mine samples. In recent years, although deep learning and reinforcement learning technologies have been gradually introduced into the field of spectral analysis, existing solutions mostly focus on single modal data processing, lack the ability to collaboratively model XRF and PAA multi-source data, and result in limited correction accuracy. In addition, when laboratory standard models are transferred to actual mine site scenarios, large errors often occur due to matrix differences, and an intelligent analysis method that can fuse multi-source spectral data, dynamically optimize correction parameters, and adapt to complex matrices is urgently needed. SUMMARY

[0003] The purpose of the present application is to provide a mine sample analysis method and system based on spectral correction, which can realize the precision, intelligence and efficiency of power service.

[0004] The present application provides a mine sample analysis method based on spectral correction, which comprises:

[0005] S1: pre-treating and multi-modal spectral detecting a first mine sample, performing calibration correction and standardization processing on the multi-modal spectral detection result to obtain a first spectral data array;

[0006] S2: performing multi-modal spectral fusion processing on the first spectral data array, and determining mineral combination analysis information according to the obtained first spectral fusion feature vector;

[0007] The S2 comprises:

[0008] S21: performing format unification and elastic energy alignment processing on the first spectral data array to obtain a first spectral fusion feature vector;

[0009] S22: inputting the first spectral fusion feature vector into a mineral combination analysis model to determine the mineral combination analysis information;

[0010] S3: matching a target mine sample according to the mineral assemblage analysis information, and setting the analysis parameters of the target mine sample as an optimal photon energy-time parameter set;

[0011] S4: performing a component analysis process on the first mine sample according to the optimal photon energy-time parameter set, generating an analysis result and saving the analysis result.

[0012] Preferably, the S1 comprises:

[0013] S11: performing a first pretreatment operation on a first mine sample to obtain a target mine sample;

[0014] S12: performing a first spectrum detection process on the target mine sample to obtain a first spectrum detection result;

[0015] S13: performing a second spectrum detection process on the target mine sample to obtain a second spectrum detection result;

[0016] S14: performing a calibration correction and standardization process on the first spectrum data and the second spectrum data to obtain a first spectrum data array.

[0017] Preferably, the first pretreatment operation comprises a crushing process, a drying process and a tabletting process.

[0018] Preferably, the S14 comprises:

[0019] S141: performing an energy calibration correction process on the first spectrum data and the second spectrum data respectively to obtain first corrected spectrum data and second corrected spectrum data;

[0020] S142: performing a background smoothing process on the first corrected spectrum data and the second corrected spectrum data respectively to obtain first target spectrum data and second target spectrum data;

[0021] S143: performing a data format standardization process on the first target spectrum data and the second target spectrum data to obtain a first spectrum data array.

[0022] Preferably, the S2 comprises:

[0023] S21: performing a format unification and elastic energy alignment process on the first spectrum data array to obtain a first spectrum fusion feature vector;

[0024] S22: inputting the first spectrum fusion feature vector into a mineral assemblage analysis model to determine mineral assemblage analysis information.

[0025] Preferably, the S21 comprises:

[0026] S211: performing format unification processing on the first spectrum data array to obtain a first spectrum fusion feature vector;

[0027] S212: performing energy axis elastic alignment processing on the second spectrum data array to obtain a target spectrum data array;

[0028] S213: obtaining a first spectrum fusion feature vector based on the target spectrum data array.

[0029] Preferably, the S3 comprises:

[0030] S31: performing similarity matching processing in a historical mine sample database by using the mineral combination analysis information to obtain at least one first target mine sample;

[0031] S32: performing double ordering processing on at least one of the first target mine samples to obtain a target mine sample and an optimal photon energy-time parameter set.

[0032] Preferably, the S4 comprises:

[0033] S41: performing component analysis processing on the first mine sample according to the optimal photon energy-time parameter set to generate a structured report;

[0034] S42: uploading metadata generated in the component analysis processing to a blockchain for storage.

[0035] Preferably, the mineral combination analysis information is in the form of a multi-element vector.

[0036] The present application also proposes a mine sample analysis system based on spectrum correction, which is used to implement the above-mentioned mine sample analysis method based on spectrum correction.

[0037] The mine sample analysis method and system based on spectrum correction proposed in the present application relate to the technical field of mine component analysis. First, a first mine sample is pretreated, and XRF spectrum and gamma spectrum are obtained respectively, then fusion processing is performed on the two spectra to obtain a first spectrum fusion feature vector, mineral combination analysis information of the first mine sample is obtained based on machine learning model analysis, finally, at least one mine sample whose similarity meets a preset requirement is found in historical mine sample data according to the mineral combination analysis information, and the collection parameters of the determined target mine sample are used as an optimal photon energy-time parameter set, analysis processing of the first mine sample is completed and an analysis report is formed. The technical solution of the present application determines the best analysis parameters through two steps of artificial intelligence and big data screening, so that the mine analysis result is more accurate. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the accompanying drawings required by the embodiments or the prior art description will be briefly introduced below. Obviously, the accompanying drawings in the following description are only exemplary, and other drawings can be obtained by the provided drawings without creative labor for those skilled in the art.

[0039] Figure 1 is an execution flow diagram of a mine sample analysis method based on spectrum correction in the present application.

[0040] Figure 2 is a schematic diagram of a pretreatment process of a mine sample in the present application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0042] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments, in which the schematic embodiments and the description are only used to explain the present application, but not as a limitation to the present application.

[0043] A mine sample analysis method and system based on spectrum correction in the present application will be described in detail below.

[0044] The present embodiment proposes a mine sample analysis method based on spectrum correction, and the specific process is as shown in Figure 1

[0045] S1: The first mine sample is pretreated and multi-modal spectrum detection is performed, and the multi-modal spectrum detection result is subjected to calibration correction and standardization processing to obtain a first spectrum data array.

[0046] In this step, the mine sample needs to be pretreated and spectrum acquisition, and the spectrum data is subjected to standardization processing, so as to prepare for the subsequent analysis and processing process.

[0047] The S1 includes the following sub-steps:

[0048] S11: The first pretreatment operation is performed on the first mine sample to obtain a target mine sample.

[0049] ​In this step, the first pretreatment operation includes a crushing process, a drying process and a tabletting process. Through the above processes, the full automation and accurate control of the sample particle size, moisture content and tablet quality can be achieved, and the detection errors caused by uneven particle size, residual moisture and sample deformation in traditional manual operation can be solved.

[0050] As shown in Figure 2 The pretreatment process of the mine sample mainly includes:

[0051] 1. Crushing process: a two-stage closed-circuit crushing process is adopted, which includes a jaw crusher, a vibrating screen and a roller crusher. The particle size distribution is monitored in real time by a PLC control system to ensure that the final particle size is less than the preset value.

[0052] 2. Drying process: an infrared moisture detection module is integrated, and a dynamic humidity compensation mechanism is set in a 105°C air drying oven. When the moisture content of the sample is greater than 0.5%, the drying time is automatically extended. For example, a gradient drying mode can be used: the initial drying time is 2 hours, and when the moisture content of the sample is detected to be less than the required value, the compensation time is increased according to the difference between the moisture content of the sample and the preset value, wherein the compensation time can be set to 0-30 minutes, and the greater the difference, the longer the compensation time.

[0053] 3. Tabletting process: a hydraulic tablet press is integrated with a three-dimensional pressure sensor and a laser flatness detection system. Through feedback control, the pressure uniformity and sample flatness error can meet the preset requirements.

[0054] S12: performing a first spectral detection process on the target mine sample to obtain a first spectral detection result.

[0055] In this step, the scanning parameters are dynamically optimized and the matrix effect is corrected, breaking through the limitations of traditional fixed parameter scanning and improving the detection accuracy of complex matrix samples.

[0056] The first spectral detection result refers to a multi-dimensional XRF energy spectrum.

[0057] The specific process of scanning is as follows:

[0058] 1. Hardware configuration: Thermo Scientific ARL 9900 XRF spectrometer is adopted, equipped with Rh target anode and SDD detector array. The configuration parameters of the Rh target anode are preferably 50kV / 60mA, and the configuration parameters of the SDD detector array are preferably 4x40mm². In order to ensure the accuracy and reliability of the detection result, the vacuum environment needs to be maintained at ≤0.1mbar.

[0059] 2. Software enhancement: Since different mine sample types are suitable for different scanning modes, it is necessary to dynamically adjust the scanning parameters according to the sample matrix type. Specifically, for high-silicon samples, step scanning is required, and the scanning step length can be selected as 100 ms / step, while for high-sulfur samples, continuous scanning mode is required.

[0060] S13: performing a second spectral detection process on the target mine sample to obtain a second spectral detection result.

[0061] In the process of mine sample analysis, the type of mine can be quickly detected by X-ray, but the content of radioactive substances in the mine still needs to be analyzed by γ-ray spectrum.

[0062] The second spectral detection result refers to the γ-ray spectrum.

[0063] In this step, through beam monitoring and wavelet-Bayesian joint algorithm, the problem of weak signal of low-abundance nuclide and high background noise is solved, and the γ-ray spectrum analysis capability is significantly improved.

[0064] The process of γ-ray spectrum analysis mainly includes:

[0065] Accelerator configuration: Varian TrueBeam linear accelerator is adopted, which integrates real-time beam monitoring to ensure that the irradiation uniformity error is within the preset range.

[0066] The beam monitor is usually located at the outlet of the particle conduit and before the sample table. It measures the incident particle intensity changing over time in real time, providing normalized parameters for the spectrometer detector, thereby effectively reducing the impact of changes in particle beam intensity.

[0067] Detector upgrade: ORTEC GEM series high-purity germanium detectors are used, which are combined with digital signal processors to realize nanosecond-level γ-ray pulse resolution, so that the energy resolution is maintained within the preset range.

[0068] Data preprocessing: noise suppression algorithm based on wavelet transform is used to remove noise, thereby significantly improving the detection sensitivity of low-abundance nuclides. The noise suppression algorithm based on wavelet transform can use known algorithms in the prior art, which will not be described here.

[0069] S14: performing calibration correction and standardization processing on the first spectral data and the second spectral data to obtain a first spectral data array.

[0070] In this step, the first spectral data and the second spectral data need to be corrected and standardized to facilitate subsequent analysis and processing.

[0071] The S14 includes the following sub-steps:

[0072] S141: Perform energy calibration correction processing on the first spectral data and the second spectral data respectively to obtain first corrected spectral data and second corrected spectral data.

[0073] In this step, a mathematical mapping relationship between channel address and energy is established by using a polynomial fitting algorithm.

[0074] The specific process includes: first, reading the standard radioactive source spectrum file, using an automatic peak finding algorithm to accurately identify the channel address position of the characteristic peak of the standard source; then, establishing a data pair set of channel address and known energy; then, using the least squares method to perform polynomial fitting on the data, and solving the fitting coefficients; finally, applying this fitting formula to all sample spectral data, converting each channel address to the corresponding energy value, and generating the energy-scaled spectrum.

[0075] S142: Perform background smoothing processing on the first corrected spectral data and the second corrected spectral data respectively to obtain first target spectral data and second target spectral data.

[0076] In this step, corresponding background smoothing processing needs to be performed on the characteristics of the first corrected spectral data and the second corrected spectral data respectively to obtain the first target spectral data and the second target spectral data.

[0077] Specifically, for XRF energy spectrum, the SNIP (Statistics-sensitive Non-linear Iterative Peak-clipping) algorithm is used, which gradually deducts peaks with counts higher than the "background window" through multiple iterations, and finally fits the background profile of the entire energy region. For PAA gamma spectrum, the adaptive iterative least squares method is used to fit the Compton continuous background. Preferably, the number of iterations of the algorithm is set to 50-100 times, and the window width is set to 15-25 channels according to the spectral line width.

[0078] S143: Perform data format standardization processing on the first target spectral data and the second target spectral data to obtain a first spectral data array.

[0079] The data format standardization processing refers to saving the first target spectral data and the second target spectral data into the first spectral data array in a specified format.

[0080] Specifically, the processed data needs to be packaged into JSON format, which is lightweight, readable and easy to be parsed by various programming languages and subsequent algorithms. The data structure includes sample number, collection time, analysis technology type, energy calibration coefficient, and spectral data array.

[0081] Preferably, each element in the first spectral data array further contains information such as energy value, count intensity, corresponding element, peak net area, and half-height width.

[0082] S2: Perform multi-modal spectral fusion processing on the first spectral data array, and determine mineral combination analysis information according to the obtained first spectral fusion feature vector.

[0083] In this step, the two types of spectral data in the first spectral data array obtained in S1 need to be fused, and the type of mineral is identified based on the fusion result to prepare for subsequent analysis.

[0084] The S2 includes the following sub-steps:

[0085] S21: Perform format unification and elastic energy alignment processing on the first spectral data array to obtain a first spectral fusion feature vector.

[0086] In order to establish an XRF-γ energy spectrum combined feature space, the XRF energy spectrum data and the γ energy spectrum data in the first spectral data array need to be elastically aligned on the energy axis, so that the obtained fusion feature vector conforms to the actual data situation.

[0087] The S21 includes the following sub-steps:

[0088] S211: Perform format unification processing on the first spectral data array to obtain a first spectral fusion feature vector.

[0089] In this step, the XRF energy spectrum data and the γ energy spectrum data are extracted from the first spectral data array, and format unification and alignment processing are performed on the two, so as to form a fusion feature vector subsequently.

[0090] The format unification processing includes:

[0091] 1. Unit unification:

[0092] Convert the energy unit (MeV) in the XRF energy spectrum to keV (1 MeV = 1000 keV), and ensure that the XRF and γ energy spectrum axis units are consistent.

[0093] 2. Energy axis interpolation:

[0094] Perform piecewise linear interpolation on the energy axis of the XRF energy spectrum and the γ energy spectrum to ensure that the energy resolution of the two is consistent.

[0095] 3. Abnormal point elimination:

[0096] Use the 3σ principle to filter outliers in the XRF / γ energy spectrum data, and the specific elimination rules can be set according to specific needs.

[0097] After this step, the unit and energy axis of the XRF spectrum data and the gamma spectrum data are kept uniform.

[0098] S212: Perform energy axis elastic alignment processing on the second spectrum data array to obtain a target spectrum data array.

[0099] In this step, an XRF-gamma spectrum combined feature space is constructed, and dynamic time warping (DTW) algorithm is used to realize energy axis elastic alignment.

[0100] Dynamic time warping is different from traditional Euclidean distance comparison. The traditional Euclidean distance comparison requires that the lengths of two sequences are equal and one-to-one correspondence. This is very inaccurate for sequences that have stretching, shifting, and speed changes on the time axis.

[0101] DTW allows one-to-many mapping of points in the sequence, thereby "flexibly" bending or warping the time axis to find a path that best matches the overall shape of the two sequences, even if their lengths are different.

[0102] After the elastic alignment processing of the XRF spectrum data and the gamma spectrum data in the second spectrum data array in this step, the target spectrum data array is obtained.

[0103] S213: Obtain a first spectrum fusion feature vector based on the target spectrum data array.

[0104] In this step, the XRF spectrum data and the gamma spectrum data in the target spectrum data array need to be fused.

[0105] In the specific fusion processing process, a feature-level fusion method is used to splice the XRF spectrum data and the gamma spectrum data to obtain the first spectrum fusion feature vector.

[0106] Preferably, in one embodiment, principal component analysis processing can be first performed on the XRF spectrum data and the gamma spectrum data to extract at least one principal component information, and then the extracted principal component information is spliced to obtain the first spectrum fusion feature vector.

[0107] S22: Input the first spectrum fusion feature vector into a mineral combination analysis model to determine mineral combination analysis information.

[0108] In this step, the first spectrum fusion feature vector obtained in S21 needs to be input into a mineral combination analysis model to determine mineral combination analysis information.

[0109] Preferably, the mineral combination analysis model can use a convolutional neural network model.

[0110] Training data: Construct a million-level ore spectrum database that can cover a sufficient variety of ore types, use SMOTE oversampling technology to balance data distribution, and ensure that the test set accuracy meets the preset requirements.

[0111] For each training data, the characteristics of the mine content and the finally determined mine type need to be labeled, such as "chalcopyrite + high sulfur -> Cu mine", so as to generate mineral combination analysis information with confidence labels.

[0112] The spectrum fusion feature vector in the training data is used as input data, and the mineral combination analysis information is used as output data to train the mineral combination analysis model.

[0113] Preferably, the mineral combination analysis information can include five-level mineralogical characteristic labels, including ore type -> element combination -> mineral type -> associated mineral, such as ["copper mine", "high sulfur", "gold containing", "galena associated"].

[0114] Preferably, the mineral combination analysis information is in the form of a multi-element vector.

[0115] S3: According to the mineral combination analysis information, a target mine sample is matched, and the analysis parameter set of the target mine sample is used as the optimal photon energy-time parameter set.

[0116] In S2, the mine sample is judged qualitatively and chemically, but this can only be used as a preliminary judgment result. In order to obtain more accurate detection results, further quantitative and physical layer judgment is required. In this step, at least one similar mineral sample can be determined based on the mineral combination analysis information, and the instrument acquisition parameters of at least one similar mineral sample can be used to realize fine component analysis of the first mine sample.

[0117] S3 includes the following sub-steps:

[0118] S31: Use the mineral combination analysis information to perform similarity matching processing in the historical mine sample database to obtain at least one first target mine sample.

[0119] The historical mine sample database stores historical mine sample data corresponding to multiple types of mines.

[0120] Preferably, the historical mine sample database uses Neo4j graph database to store historical records, constructs a three-element relationship graph (Sample-Parameter-Performance), and supports complex queries (such as "parameter set of gold-containing copper mine with Au detection limit <1ppm").

[0121] The similarity matching process adopts a weighted cosine similarity calculation method.

[0122] S32: Perform double-ordering processing on at least one of the first target mine samples to obtain a target mine sample and an optimal photon energy-time parameter set.

[0123] In this step, it is necessary to perform an optimal parameter set decision processing on at least one of the first target mine samples to select at least one of the first target mine samples that is most similar to the mine composition of the first mine sample, so as to determine the optimal photon energy-time parameter set.

[0124] Preferably, all the first target mine samples retrieved are sorted according to the data quality indicators stored therein by means of evaluation index sorting. After obtaining the sorting result, the first N target mine samples are taken as second target mine samples, and the first M second target mine samples are taken as target mine samples according to the analysis accuracy and detection limit.

[0125] Preferably, N is 5 and M is 1.

[0126] After the target mine sample is determined, the system automatically obtains the instrument parameters used thereby to generate an optimal photon energy-time parameter set dedicated to the current sample. The content thereof is a structured key parameter, such as

[0127] {

[0128] "sample_type": "high-sulfur copper-gold mine",

[0129] "recommended_energy_range_mev": [13.5, 15.0],

[0130] "recommended_irradiation_time_s": 90,

[0131] "target_elements": ["Au", "Ag", "Cu"],

[0132] "source_record_id": "DB_Record_#12345",

[0133] "expected_improvement": "Au detection limit expected to be reduced by 40%"

[0134] }。

[0135] S4: performing component analysis on the first mine sample according to the optimal photon energy-time parameter set to generate an analysis result and save the analysis result.

[0136] In this step, the first mine sample can be subjected to refined component analysis according to the optimal photon energy-time parameter set to output the result.

[0137] The S4 includes the following sub-steps:

[0138] S41: performing component analysis on the first mine sample according to the optimal photon energy-time parameter set to generate a structured report.

[0139] The structured report contains the content of target elements (such as Au: 5.2 g / t, Ag: 83 g / t) and a spectral visualization map.

[0140] The component analysis process can refer to the known mine component analysis process in the prior art.

[0141] S42: uploading metadata generated in the component analysis process to a blockchain for saving.

[0142] In this step, the metadata (such as device parameters) of the analysis process is recorded by the blockchain technology to ensure data traceability.

[0143] The present application also proposes a mine sample analysis system based on spectral correction for performing the above-mentioned mine sample analysis method based on spectral correction.

[0144] The mine sample analysis method and system based on spectral correction proposed in the present application relate to the technical field of mine component analysis. First, a first mine sample is pretreated, and XRF energy spectrum and gamma energy spectrum are obtained respectively, then the two energy spectrums are fused to obtain a first spectral fusion feature vector, the mineral combination analysis information of the first mine sample is obtained based on a machine learning model analysis, finally, at least one mine sample with a similarity satisfying a preset requirement is found in historical mine sample data according to the mineral combination analysis information, and the collection parameters of the determined target mine sample are taken as the optimal photon energy-time parameter set, the analysis processing of the first mine sample is completed, and an analysis report is formed. The technical solution of the present application determines the best analysis parameters through two steps of artificial intelligence and big data screening, so that the mine analysis result is more accurate.

[0145] The above only describes the preferred embodiments of the present application, and any equivalent changes or modifications made to the structure, features and principles described in the scope of the present application are included in the scope of the present application.

Claims

1. A method of mine sample analysis based on spectral correction, characterized by, The method comprises: S1: performing pretreatment and multi-modal spectrum detection on a first mine sample, performing scale correction and standardization processing on the multi-modal spectrum detection result to obtain a first spectrum data array; S2: performing multi-modal spectrum fusion processing on the first spectrum data array, and determining mine combination analysis information according to the obtained first spectrum fusion feature vector; The S2 comprises: S21: performing format unification and elastic energy alignment processing on the first spectrum data array to obtain a first spectrum fusion feature vector; S22: inputting the first spectrum fusion feature vector into a mine combination analysis model to determine mine combination analysis information; S3: matching a target mine sample according to the mine combination analysis information, and taking a set of analysis parameters of the target mine sample as an optimal photon energy-time parameter set; The S3 comprises: S31: performing similarity matching processing in a historical mine sample database by using the mine combination analysis information to obtain at least one first target mine sample; S32: performing double sorting processing on at least one first target mine sample to obtain a target mine sample and an optimal photon energy-time parameter set; S4: performing component analysis processing on the first mine sample according to the optimal photon energy-time parameter set, generating an analysis result and saving the analysis result; The S4 comprises: S41: performing component analysis processing on the first mine sample according to the optimal photon energy-time parameter set to generate a structured report; S42: uploading metadata generated in the component analysis processing to a blockchain for saving.

2. A mine sample analysis method based on spectral correction according to claim 1, characterized in that, The S1 comprises: S11: performing a first pretreatment operation on a first mine sample to obtain a target mine sample; S12: performing first spectrum detection processing on the target mine sample to obtain a first spectrum detection result; S13: performing second spectrum detection processing on the target mine sample to obtain a second spectrum detection result; S14: performing scale correction and standardization processing on the first spectrum detection result and the second spectrum detection result to obtain a first spectrum data array.

3. A mine sample analysis method based on spectral correction according to claim 2, characterized in that, The first pretreatment operation comprises a crushing process, a drying process and a tabletting process.

4. A mine sample analysis method based on spectral correction according to claim 2, characterized in that, The S14 comprises: S141: performing energy scale correction processing on the first spectrum detection result and the second spectrum detection result respectively to obtain first corrected spectrum data and second corrected spectrum data; S142: performing background smoothing processing on the first corrected spectrum data and the second corrected spectrum data respectively to obtain first target spectrum data and second target spectrum data; S143: performing data format standardization processing on the first target spectrum data and the second target spectrum data to obtain a first spectrum data array.

5. The mine sample analysis method based on spectral correction according to claim 1, characterized in that, The S21 comprises: S211: performing format unification processing on the first spectrum data array to obtain a second spectrum data array; S212: performing energy axis elastic alignment processing on the second spectrum data array to obtain a target spectrum data array; S213: obtaining a first spectrum fusion feature vector based on the target spectrum data array.

6. A mine sample analysis method based on spectral correction according to claim 1, characterized in that, The mineralogical analysis information is in the form of a multivariate vector.

7. A mine sample analysis system based on spectral correction, characterized by, A method for implementing a spectral correction-based mine sample analysis according to any one of claims 1-6.

Images