A remote sensing spectrum inversion method and system for tailings composition

Abstract

The embodiment of the present application provides a method and system for remote sensing spectral inversion of tailings components, the method can obtain remote sensing spectral data and laboratory spectral data. And based on the direct standardization algorithm, an error transfer function between laboratory spectral data and remote sensing spectral data is constructed. The error transfer function is then used to perform correction on the remote sensing spectral data to obtain corrected spectral data. Then, according to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated. And a regression model is constructed based on the tailings spectral index, and the regression model is used to estimate the main component content corresponding to the corrected spectral data. The method can correct the remote sensing spectral data through the error transfer function, so that the regression model constructed based on the laboratory spectral data can be directly applied to the remote sensing spectral data, estimate the main component content of the tailings, and improve the accuracy of detecting the main component content of the tailings based on the remote sensing spectral data.

Description

Technical Field

[0001] The present application relates to the technical field of spectral component analysis, and in particular to a method and system for remote sensing spectral inversion of tailings components. Background Art

[0002] Tailings refers to the waste discharged after the ore is ground and the useful components are selected before mineral smelting. Tailings can be solid mineral waste formed by dehydrating the tailings slurry discharged by the beneficiation plant. Taking the iron ore smelting process as an example, the production of 1 ton of iron ore concentrate will produce 2.5 to 3.0 tons of iron tailings. The fine-grained tailings and wastewater after grinding and beneficiation are discharged in the form of slurry and stored in the tailings pond. The accumulation of a large amount of iron tailings will bring dust, harm the surrounding environment, and there is a risk of dam breach.

[0003] The main components of tailings may include metal compounds, SiO2, etc. By effectively monitoring the main components of tailings, tailings can be recycled and reused according to the components, which is beneficial to the safe management of tailings ponds and the protection of the surrounding environment. In order to detect the main components in tailings, point sampling can be carried out and chemical tests can be performed on the tailings samples obtained by sampling. This detection method requires a large amount of time and labor costs, and repeated monitoring is not possible.

[0004] The main components of tailings can also be detected through remote sensing spectroscopy. After performing spectral scanning on tailings samples through remote sensing equipment to obtain remote sensing spectral data, the remote sensing spectral data is analyzed to obtain parameters such as reflectivity of each band. The main components are determined based on parameters such as reflectivity. However, due to the influence of various factors such as atmospheric radiation transmission and tailings water content, the tailings reflectivity spectrum in remote sensing spectral data is quite different from the laboratory spectrum, resulting in inaccurate component detection results when the tailings attribute inversion model constructed based on laboratory reflectivity spectrum is applied to remote sensing spectral data. Summary of the invention

[0005] In view of this, an embodiment of the present application provides a method and system for remote sensing spectral inversion of tailings components to improve the accuracy of remote sensing spectral detection of main components.

[0006] According to one aspect of the present application, a method for remote sensing spectral inversion of tailings components is provided, the method comprising:

[0007] Acquire spectral data, wherein the spectral data includes remote sensing spectral data and laboratory spectral data; the remote sensing spectral data is data obtained by performing spectral detection on a target area through a remote sensing device; the laboratory spectral data is data obtained by performing spectral detection on a tailings sample collected in the target area through a spectrometer;

[0008] constructing an error transfer function between the laboratory spectral data and the remote sensing spectral data based on a direct normalization algorithm;

[0009] Using the error transfer function to perform correction on the remote sensing spectral data to obtain corrected spectral data;

[0010] According to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated, wherein the training spectral data is data randomly extracted from the laboratory spectral data according to a preset ratio; the tailings spectral index is the band ratio between the tailings attribute sensitive band and the tailings attribute insensitive band of the main component in the training spectral data;

[0011] A regression model is constructed according to the tailings spectral index, and the regression model is used to estimate the main component content corresponding to the corrected spectral data.

[0012] Optionally, the method further includes:

[0013] Obtaining a first sample weight, where the first sample weight is the weight weighed when the sampling container is filled with the wet tailings sample;

[0014] Obtaining a second sample weight, where the second sample weight is the weight weighed when the sampling container is filled with a dry tailings sample; the dry tailings sample is obtained by drying the wet tailings sample filled in the sampling container;

[0015] Obtain the weight of the sampling container;

[0016] Calculating a first difference and a second difference, wherein the first difference is the difference between the weight of the first sample and the weight of the second sample; and the second difference is the difference between the weight of the second sample and the weight of the sampling container;

[0017] The sample moisture content is calculated according to the first difference and the second difference. The sample moisture content is equal to the ratio of the first difference to the second difference. The sample moisture content is used to collect tailings samples with different moisture contents.

[0018] Optionally, the method further includes:

[0019] Set up X-ray diffraction measurement parameters;

[0020] controlling the X-ray diffraction device to emit X-rays to the ground dry tailings sample based on the X-ray diffraction measurement parameters;

[0021] Acquiring diffraction data, wherein the diffraction data includes the position, diffraction intensity and diffraction peak width of the diffraction peak;

[0022] The main components of the tailings sample are determined based on the diffraction data.

[0023] Optionally, obtain spectral data, including:

[0024] Set the spectrum measurement parameters;

[0025] Controlling the spectrometer to emit measurement light to the dry tailings sample according to the spectral measurement parameters, wherein the spectrometer performs whiteboard calibration according to a preset calibration cycle during the measurement process;

[0026] receiving at least four pieces of spectral data detected by the reflectance spectrum sensor of the spectrometer; the at least four pieces of spectral data are respectively reflectivities detected by the reflectance spectrum sensor after the dry tailings sample is rotated;

[0027] An average value of at least four pieces of the spectral data is calculated to obtain laboratory spectral data of the tailings sample.

[0028] Optionally, the method further includes:

[0029] performing a smoothing process on the laboratory spectral data;

[0030] Dividing the smoothed laboratory spectral data into a training set and a validation set, with the ratio of the training set to the validation set being 3:1;

[0031] Acquiring the spatial resolution of the remote sensing spectral data;

[0032] The laboratory spectral data is resampled according to the spatial resolution to obtain the training spectral data.

[0033] Optionally, according to a multi-objective planning algorithm, a tailings spectral index of the training spectral data is calculated, including:

[0034] extracting potential band combinations from the training spectral data;

[0035] defining a decision space of a multi-objective programming algorithm, wherein the decision space includes all band combinations within the reflectance spectral absorption feature range;

[0036] defining a target function of the content of the main components in the tailings sample;

[0037] Setting a cost function for band selection screening according to the content objective function;

[0038] Based on the cost function, the potential band combination and the decision space, calculating the band ratio between the tailings property sensitive band and the tailings property insensitive band in the training spectral data;

[0039] An optimal band combination is determined according to the band ratio.

[0040] Optionally, according to a multi-objective planning algorithm, a tailings spectral index of the training spectral data is calculated, including:

[0041] Extracting a characteristic band from the training spectrum data, wherein the characteristic band is a band where absorption characteristics appear;

[0042] Using a multi-objective programming algorithm, an optimal band combination of target substance content is obtained, wherein the target substance is one of the main components of the tailings sample; the optimal band combination includes a wavelength value determined in the characteristic band according to the multi-objective programming algorithm;

[0043] Obtaining correlation information between the reflectance spectrum and tailings properties, wherein the correlation information is used to characterize a mapping relationship between wavelength and correlation coefficient;

[0044] Determining the sensitive wavelength and the insensitive wavelength of the target substance according to the correlation information;

[0045] A tailings spectral index is calculated according to the sensitive wavelength and the insensitive wavelength, and the tailings spectral index includes a ratio index and a three-band spectral index.

[0046] Optionally, constructing a regression model according to the tailings spectral index includes:

[0047] Establishing a regression model between the spectral index and the content of the main component based on the training set, wherein the regression model includes a linear regression model and a nonlinear regression model;

[0048] Performing accuracy verification on the regression model using the verification set;

[0049] The regression model validation accuracy is obtained, wherein the regression model validation accuracy includes selecting a determination coefficient and a root mean square error.

[0050] Optionally, the method further includes:

[0051] Get the accuracy evaluation interval;

[0052] If the regression model verification accuracy is within the accuracy evaluation interval, plotting a spatial distribution map of the main components based on the regression model and the corrected spectral data;

[0053] If the regression model verification accuracy is not within the accuracy evaluation interval, a three-band spectral index is obtained, and the regression model is reconstructed based on the three-band spectral index.

[0054] According to another aspect of the present application, a tailings composition remote sensing spectral inversion system is provided, the system comprising:

[0055] A spectrum acquisition module is used to acquire spectrum data, wherein the spectrum data includes remote sensing spectrum data and laboratory spectrum data; the remote sensing spectrum data is data obtained by performing spectrum detection on a target area through a remote sensing device; the laboratory spectrum data is data obtained by performing spectrum detection on a tailings sample collected in the target area through a spectrometer;

[0056] A transfer function construction module, used for constructing an error transfer function between the laboratory spectral data and the remote sensing spectral data based on a direct standardization algorithm;

[0057] A correction module, used for correcting the remote sensing spectral data using the error transfer function to obtain corrected spectral data;

[0058] A planning module, for calculating, according to a multi-objective planning algorithm, a tailings spectral index of training spectral data, wherein the training spectral data is data randomly extracted from the laboratory spectral data according to a preset ratio; the tailings spectral index is a band ratio between a tailings attribute sensitive band and a tailings attribute insensitive band in the training spectral data;

[0059] The regression analysis module is used to construct a regression model according to the tailings spectral index, and use the regression model to estimate the main component content corresponding to the corrected spectral data.

[0060] By means of the above technical scheme, a method and system for remote sensing spectral inversion of tailings components provided in an embodiment of the present application can obtain remote sensing spectral data and laboratory spectral data. And an error transfer function between laboratory spectral data and remote sensing spectral data is constructed based on a direct standardization algorithm. The remote sensing spectral data is then corrected using the error transfer function to obtain corrected spectral data. Then, according to a multi-objective planning algorithm, the tailings spectral index of the spectral data is calculated. And a regression model is constructed based on the tailings spectral index, and the regression model is used to estimate the main component content corresponding to the corrected spectral data. The method can correct the remote sensing spectral data through an error transfer function, so that the regression model constructed based on laboratory spectral data can be directly applied to remote sensing spectral data, estimate the main component content of tailings, and improve the accuracy of detecting the main component content of tailings based on remote sensing spectral data.

[0061] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0063] Figure 1 A schematic diagram of a process for remote sensing spectral inversion of tailings components provided in an embodiment of the present application is shown;

[0064] Figure 2 A schematic diagram of the main component flow of X-ray diffraction detection provided in an embodiment of the present application is shown;

[0065] Figure 3 A schematic diagram of the spectrum detection process provided by an embodiment of the present application is shown;

[0066] Figure 4 A schematic diagram of the process of dividing the training set and the validation set provided in an embodiment of the present application is shown;

[0067] Figure 5 A schematic diagram of a multi-objective planning algorithm flow chart provided in an embodiment of the present application is shown;

[0068] Figure 6 A schematic diagram of reflectivity detection results provided by an embodiment of the present application is shown;

[0069] Figure 7 A schematic diagram of the relationship between wavelength correlation coefficients provided in an embodiment of the present application is shown;

[0070] Figure 8 A schematic diagram of TFe spectral index provided in an embodiment of the present application is shown;

[0071] Fig. 9 A schematic diagram of SiO2 spectral index provided in an embodiment of the present application is shown;

[0072] Fig.10 A schematic diagram of the accuracy evaluation process provided by an embodiment of the present application is shown;

[0073] Fig.11 The spatial distribution diagram of the tailings pond TBI provided in the embodiment of the present application is shown;

[0074] Fig.12 A schematic diagram of TBI with different particle size grades and water contents of TFe and SiO2 provided in the embodiments of the present application is shown;

[0075] Fig.13 A schematic structural diagram of a tailings composition remote sensing spectral inversion system provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0076] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0077] In an embodiment of the present application, the tailings composition remote sensing spectral inversion method can be applied to mineral smelting process. Mineral smelting process is the process of converting mineral raw materials into metals or metal compounds through a series of physical and chemical processes. In the mineral smelting process, mineral smelting needs to grind the ore first and select useful components. The waste discharged after the useful components are selected is tailings. Therefore, tailings is a kind of ore waste generated in the mineral smelting process.

[0078] In some embodiments, tailings may be a solid mineral waste with a fine particle structure, formed by dehydrating the tailings slurry discharged from the ore dressing plant. Tailings may include various types according to the final material obtained by the mineral smelting process. For example, the tailings produced by the steel smelting process are iron tailings, and the tailings produced by the aluminum smelting process are aluminum tailings.

[0079] The tailings may include compounds containing the target element and impurity components. Among them, the target element refers to the final material element obtained by the mineral smelting process. That is, for the steel smelting process, the target element is the iron (Fe) element. In the tailings, the target element usually exists in the form of compounds, such as Fe2O3, Fe3O4, FeCO3, ferric hydroxide, FeTiO3, FeS2, etc. These compounds contain target element components and can be collectively referred to as compounds containing the target element. For ease of description, in the embodiments of the present application, the target element compound is represented by the form of "T+target element", for example, the iron-containing compound can be represented as TFe.

[0080] Impurity components are the associated substances of minerals. The impurity components contained in minerals vary depending on the mining area. For example, for iron ore, the impurity components are mainly SiO2, Al2O3, CaO, etc. It can be seen that the target element compounds and some impurity components in the tailings have certain utilization value. The high-content usable components can be used by recycling the tailings. It is also possible to determine whether the tailings constitute environmental pollution by detecting the main components in the tailings, so as to carry out appropriate tailings treatment methods.

[0081] In some embodiments, the main components of tailings can be tested by destructive testing and non-destructive testing. Among them, destructive testing refers to the component testing of tailings samples by chemical reagents. Since chemical reagents do not have on-site operation conditions, destructive testing requires random sampling or fixed-point sampling on-site, and the tailings samples are sent to a designated laboratory for testing. The component results of this detection method are highly accurate, but the time and labor costs are very high, and repeated monitoring cannot be performed.

[0082] Nondestructive testing refers to the detection of the composition of tailings samples by means of X-ray diffraction, spectral detection, etc. Taking the spectral detection process as an example, the spectrometer needs to be initially set up before the test to ensure that the settings of the spectrometer match the needs of the experiment. Then select the appropriate sample form according to the detection requirements. For tailings detection, the sample form is a powdered solid. Then, according to the characteristics of the sample, adjust the position and setting angle of the light source and detector, ensure that the light source and detector are aligned, and adjust the focus of the incident light beam to obtain the best signal intensity. Then start the spectrometer and start measuring. During the measurement process, the light reflectance and intensity at different wavelengths can be recorded to obtain spectral data. By analyzing the measured spectral data, such as drawing curves, calculating peak positions, calculating integrated intensity, etc., the main components of the tailings can be estimated.

[0083] In some embodiments, when performing spectral detection, different detection bands of light can be used for detection of the main components in the tailings. For example, the detection band of the spectrometer may include a visible light band (covering a wavelength range of 300nm to 700nm), an ultraviolet band (covering a wavelength range of 200nm to 400nm), a near infrared band (covering a wavelength range of 900nm to 1700nm), a visible-near infrared band (covering a wavelength range of 300nm to 1700nm), a short-wave-near infrared band (covering a wavelength range of 200nm to 2500nm), an infrared light band (covering a wavelength range of 700nm to 1mm), an x-ray band (covering a wavelength range of 0.01nm to 10nm), and the like.

[0084] In the embodiments of the present application, a visible near-infrared spectrometer is used as an example to describe the spectrum detection process in the tailings component remote sensing spectrum inversion method. It should be understood that the tailings component remote sensing spectrum inversion method is also applicable to the detection process of spectrometers with other detection bands, and other implementation methods that can be associated with by those skilled in the art based on the present application also belong to the protection scope of the present application.

[0085] Different substances contained in tailings can absorb light waves in different bands. Therefore, the tailings sample is measured by a spectrometer, and the reflectivity of the tailings sample in different bands can be used to calculate the content of specific substances in the tailings. For example, the reflectivity spectrum of iron tailings will show absorption characteristics in the 750-830nm, 900-1400nm and 2200-2400nm bands. The main reason for the absorption characteristics is that the iron-containing compounds (TFe) and silicon dioxide (SiO2) in the iron tailings sample absorb the spectral detection signal. Since the absorption of the main components of the tailings on the spectral detection light signal is correlated with the content of the component in the tailings, the content of the component can be inferred by analyzing the spectral detection results.

[0086] Due to the presence of Fe in minerals 3+ This results in absorption features in the reflectivity spectrum curve at 450nm, 500nm and 900nm. 2+ The presence of leads to absorption features in the reflectivity spectrum curve in the range of 900-1200nm, the absorption features in the range of 1700-1850nm can be attributed to the bending, translation and rotation combination of HOH / OH, and the absorption features in the range of 2100-2500nm can be attributed to the combination of H2O- and OH.

[0087] Based on the absorption characteristic bands of the target elements in the minerals, the visible and near-infrared (VIS-NIR) reflectance spectrum can be used to detect the main components of tailings in the metallurgical field such as iron tailings. Visible and Near-Infrared reflectance spectrum detection has the advantages of being fast, efficient, and non-destructive, and can provide a fast, cost-effective and efficient detection method for monitoring the properties of iron tailings.

[0088] For the spectral detection process, according to the different spectral data collection processes, it can include laboratory spectral detection and remote sensing spectral detection. In some embodiments, laboratory spectral detection refers to spectral detection through a spectrometer in a laboratory. When detecting the main components of tailings, laboratory spectral detection requires collecting tailings samples at the tailings discharge site, and bringing the tailings samples back to the laboratory for spectral detection through a spectrometer. This detection method has a simple detection process and low detection cost, but because the samples targeted by the detection process are obtained through random or fixed-point sampling, they are affected to a certain extent by the sampling process, making the test results specific, that is, they cannot fully reflect the content of the main components in the tailings.

[0089] In order to achieve comprehensive detection, in some embodiments, remote sensing spectrum detection can also be performed. That is, remote sensing spectrum detection is performed on the tailings discharge site through remote sensing spectrum equipment such as remote sensing satellites and remote sensing aircraft. By performing on-site detection of the tailings discharge area through remote sensing equipment, spectral data covering a large area can be obtained, and the spectral detection results in the spectral data correspond to the geographical location, thereby facilitating the analysis of a more comprehensive main component content in the discharge area.

[0090] For example, the remote sensing spectral equipment can be a high-resolution earth observation system remote sensing satellite with spectral detection function such as GF-5. The remote sensing satellite can carry an Advanced Hyperspectral Imager (AHSI) to collect hyperspectral data. The remote sensing satellite has 330 spectral bands with a spectral range from 400nm to 2500nm, and a spectral resolution of 5nm in visible and near-infrared (VNIR) and 10nm in short-wave infrared (SWIR).

[0091] The tailings component detection method based on spectral detection can use the high-dimensional information of spectral data for data analysis. The analysis process may lead to data redundancy, and the band combination algorithm (i.e., spectral index) can integrate the various bands to eliminate redundant spectral information. Among them, the optimal band combination of the spectral index can be determined by traversing the Pearson correlation coefficient between the full spectrum and the given attribute. However, this combination may destroy the physical mechanism between the reflectance spectrum and the given characteristic. In addition, due to the influence of various factors such as atmospheric radiation transmission and tailings water content, the tailings reflectance spectrum in remote sensing spectral data is quite different from the laboratory spectrum. The tailings attribute inversion model constructed based on the laboratory reflectance spectrum is difficult to be directly applied to remote sensing spectral data, resulting in inaccurate component detection results.

[0092] In order to improve the accuracy of component detection results, a tailings component remote sensing spectral inversion method is provided in this embodiment. Figure 1 As shown, the method includes:

[0093] S100: Acquire spectral data.

[0094] Wherein, the spectral data includes remote sensing spectral data and laboratory spectral data. Remote sensing spectral data is image data obtained by performing spectral detection on the target area through remote sensing equipment. The target area is the area where tailings are piled, including tailings discharge area, tailings storage area, tailings processing area, etc. The remote sensing equipment includes a spectral imager, which can perform spectral detection on the target area. For example, the remote sensing equipment can be a remote sensing satellite, an airplane, a hot air balloon, an unmanned aerial vehicle, or other aircraft carrying a spectral imager. The remote sensing equipment can perform spectral detection on the tailings discharge area from above the tailings discharge area.

[0095] Remote sensing spectral data can be acquired in real time through remote sensing equipment, or it can be extracted from a database associated with the remote sensing equipment. In some embodiments, a high-resolution satellite with remote sensing detection capabilities can store data detected during operation to a natural resources satellite remote sensing cloud service platform. When performing data analysis, data processing devices such as clients or servers can access the natural resources satellite remote sensing cloud service platform, and download remote sensing spectral data from the natural resources satellite remote sensing cloud service platform after screening and configuration as needed. For example, in order to detect remote sensing spectral data on September 2, 2023, you can access the natural resources satellite remote sensing cloud service platform and set the data time to obtain spectral image data collected by the Gaofen-5 satellite on September 2, 2023. The acquired spectral image data can retain a total of 295 bands of reflectance data after pre-processing such as radiation calibration and atmospheric correction.

[0096] Laboratory spectral data is data obtained by performing spectral detection on tailings samples collected in the target area through a spectrometer. The target area is the same as the target area for spectral detection by remote sensing equipment, and may also include tailings discharge areas, tailings storage areas, tailings treatment areas, and tailings pond areas. Tailings samples can be collected from the target area based on random sampling or fixed-point sampling.

[0097] In order to obtain relatively accurate test results, the tailings samples need to meet certain sample quantity requirements. For example, 230 surface iron tailings samples can be collected using a 100 ml capacity ring knife in 11 tailings pond areas in the iron tailings belt of AB City.

[0098] Since the water content in the tailings will affect the test results, the water content of the tailings sample can be tested when the tailings sample is collected. In some embodiments, the water content test can be performed based on air drying method, drying method, water injection method, standard compaction test, rapid consolidation test and other methods. Taking the drying method as an example, when performing the water content test, the first sample weight can be obtained first, and the first sample weight is the weight weighed when the sampling container is filled with the wet tailings sample, that is, the weight W1 of the wet tailings sample and the sampling container. Then the second sample weight is obtained, and the second sample weight is the weight weighed when the sampling container is filled with the dry tailings sample, that is, the weight W2 of the dry tailings sample and the sampling container. Among them, the dry tailings sample is obtained by drying the wet tailings sample filled in the sampling container. Then the sampling container weight W3 is obtained, and the first difference and the second difference are calculated. The first difference is the difference between the first sample weight and the second sample weight (W1-W2); the second difference is the difference between the second sample weight and the sampling container weight (W2-W3). The sample moisture content is then calculated based on the first difference and the second difference. The sample moisture content is equal to the ratio of the first difference to the second difference, that is, the moisture content is (W1-W2) / (W2-W3)×100%. The calculated sample moisture content can be used to collect tailings samples with different moisture contents, so as to construct a regression model based on the spectral data corresponding to the tailings samples with various moisture contents, thereby improving the applicability of the regression model.

[0099] After collecting the tailings samples, all tailings samples can be sealed with plastic wrap, fresh-keeping bags and other sealed containers, and then transported back to the laboratory for spectral detection by laboratory spectrometer to obtain laboratory spectral data. When performing spectral data detection, the tailings samples need to be dried first to reduce the impact of water content on subsequent test results. For example, all samples are placed in a drying box in the laboratory, the temperature is set to 105°C, and dried for about 24 hours.

[0100] After drying to obtain dry tailings samples, the main components in the dry tailings samples can be detected to determine the spectral data bands that need to be paid attention to when performing spectral detection. The main component detection can be based on the X-ray diffraction measurement method. Figure 2 As shown, in some embodiments, the X-ray diffraction measurement parameters can be set first, and then the X-ray diffraction device can be controlled to emit X-rays to the ground dry tailings sample based on the X-ray diffraction measurement parameters. Diffraction data is obtained, wherein the diffraction data includes parameters such as the position of the diffraction peak, the diffraction intensity, and the diffraction peak width used to reflect the content of the main components of the tailings sample. The main components of the tailings sample are thus determined according to the parameters in the diffraction data.

[0101] For example, after grinding the dry tailings sample and performing X-ray diffraction, and collecting the diffraction data, the phases present in the sample can be determined by comparing the diffraction pattern of the unknown sample with the standard diffraction pattern of the known substance. Since each material crystal has its own specific crystal structure parameters, the X-ray Diffraction (XRD) spectrum can reflect the specific diffraction position and diffraction intensity of the crystal characteristics, and then analyze the substance. In addition, the intensity of the diffraction line of each phase increases with the increase of its phase content. By measuring the intensity of the diffraction peak, quantitative analysis can be performed to determine the content of each phase in the sample. The main chemical components of the iron tailings sample can be determined by X-ray diffraction. It can be determined that the main chemical components of the iron tailings sample are TFe and SiO2. In the subsequent spectral detection process, the influence bands of TFe and SiO2 can be focused on, so as to set the detection parameters of the spectrometer and improve the analysis efficiency.

[0102] like Figure 3 As shown, in some embodiments, in order to obtain laboratory spectral data, when performing spectral detection, the spectral measurement parameters can be set first, and then the spectrometer can be controlled to emit measurement light to the dry tailings sample according to the spectral measurement parameters. For example, in a laboratory darkroom environment, the VIS-NIR spectrum of the dry tailings sample is measured. A 50W halogen lamp is used as the only light source, and the zenith angle is 60° to illuminate the surface of the iron tailings sample. The halogen lamp is 83cm away from the sample. The spectrometer uses a 4° field of view lens, and the sensor probe is located 55cm vertically above the tailings sample to ensure that the sensor receives all signals that are the reflection spectrum of the iron tailings.

[0103] During the measurement process, the spectrometer can be calibrated with a white board according to a preset calibration cycle. White board calibration can eliminate or reduce systematic errors caused by factors such as instrument settings, light source changes, or environmental conditions of the spectrometer, ensuring the accuracy and repeatability of spectral data. For example, white board calibration is performed every 10 minutes during the measurement process.

[0104] During the spectral detection process, the placement angle of the tailings sample also needs to be rotated so that the incident light of the full band can hit the effective image plane of the detector, thereby improving the measurement accuracy of the spectral data. For example, if the rotation angle of the tailings sample is 90 degrees, it can be rotated four times to rotate the sample 360 ​​degrees as a whole. Then receive at least 4 spectral data detected by the reflective spectrum sensor of the spectrometer. Among them, at least 4 of the spectral data are respectively the reflectance detected by the reflective spectrum sensor after the dry tailings sample is rotated. After detecting and obtaining at least 4 spectral data, the average value of the reflectance in the at least 4 spectral data can be calculated to obtain the laboratory spectral data of the tailings sample.

[0105] S200: constructing an error transfer function between the laboratory spectral data and the remote sensing spectral data based on a direct standardization algorithm.

[0106] After obtaining remote sensing spectral data and laboratory spectral data, an error transfer function can be constructed based on the direct standardization algorithm (DS). The error transfer function can be used to represent the change law between laboratory spectral data and remote sensing spectral data, thereby determining the mapping relationship between remote sensing spectral data and laboratory spectral data, and using the DS algorithm to eliminate the influence of uncertain environmental factors on remote sensing spectral data.

[0107] In some embodiments, when constructing the error transfer function, the error transfer function can be constructed by X Lab and X GF-5 They represent the spectral subset measured in the laboratory (laboratory spectral data) and the spectral subset of the tailings pond surface extracted from remote sensing satellite images (remote sensing spectral data). X Lab and X GF-5 The data size is m × n ,in m represents the number of transmission spectra, n represents the number of bands (295 can be selected in this embodiment), then the DS transmission relationship between the image spectrum data and the laboratory spectrum, that is, the error transfer function can be expressed as:

[0108]

[0109] in, D The size is n × n The parameter transfer matrix is ​​used to explain X Lab and X GF-5 The changing rules between them. R is the residual matrix, expressed as follows:

[0110]

[0111] in, r s is caused by the baseline difference n ×1-dimensional matrix, λ is a m ×1-dimensional unit column vector, all elements are 1. Therefore, the error transfer function can also be expressed as:

[0112]

[0113] In order to calculate the unknown parameter transfer matrix D First, we need to define a size of m × m The mean center matrix C m , so that:

[0114]

[0115] in, I m is a m × m dimensional identity matrix, multiply both sides of formula (3) by the mean center matrix C m ,because C m λr s T =0, C m X Lab and C m X GF-5 Defined as X Lab and X GF-5 The average central spectrum of and Therefore, formula (3) can be written as:

[0116]

[0117] From this we can get:

[0118]

[0119] in, express X GF-5 The generalized inverse matrix of . Get the parameter transfer matrix D After that, the residual matrix can be derived R , that is, according to formula (2) and formula (3), we can get:

[0120]

[0121] in, x Lab and x GF-5 It is two 1× n The row vectors of XLab and X GF-5 The average column element composition of B represents the coefficient matrix.

[0122] S300: Correct the remote sensing spectral data using the error transfer function to obtain corrected spectral data.

[0123] After the parameter transfer matrix and the residual matrix are calculated, the remote sensing spectral data can be standardized using the error transfer function, that is, the remote sensing spectral data is corrected using the error transfer function to obtain corrected spectral data:

[0124]

[0125] in, Represents remote sensing spectral data standardized based on laboratory spectral data. D represents the transfer matrix, R represents the residual matrix.

[0126] In some embodiments, in order to verify whether the spectral data obtained after the remote sensing spectral data is corrected using the error transfer function can correctly reflect the spectral detection results, the spectral angle cosine (SAC) can also be used to evaluate the similarity between the reflectance spectrum of the iron tailings sample in the corrected remote sensing spectral data and the laboratory measurement spectrum. Among them, the spectral angle cosine measures the similarity of two spectral vectors by calculating the angle between them. The smaller the calculated angle, the more similar the two spectra are, and the greater the possibility that they belong to the same type of target. The spectral angle cosine can be calculated according to the following formula:

[0127]

[0128] In the formula, cos ( x , y ) represents the cosine value between vector x and vector y, which is used to measure the similarity between two vectors in direction. or Represents the dot product (inner product) of vector x and vector y, which is used to represent the sum of the products of corresponding elements of two vectors. x 1 y 1+ x 2 y 2+…+ x n y n . and They represent the magnitude (or norm) of vectors x and y, that is, the length of the vectors.

[0129] For vector x The model is , which is the square root of the sum of the squares of the elements of the vector. and Represents vectors x and y The sum of the squares of all elements in .

[0130] The value of cosine similarity is between -1 and 1, 1 means the vectors are exactly the same, -1 means they are completely opposite, and 0 means there is no correlation between the two vectors. In the application of spectral angle cosine, cosine similarity can be used to measure the similarity between two spectra, and then used to evaluate the similarity between the reflectance spectrum of tailings samples in the corrected remote sensing spectral data and the laboratory measurement spectral data.

[0131] S400: Calculate the tailings spectral index of the training spectral data according to a multi-objective planning algorithm.

[0132] After obtaining the spectral data, the spectral data can also be preprocessed, such as Figure 4 As shown, in some embodiments, the laboratory spectral data may be smoothed first. Smoothing may reduce noise and improve data quality. The laboratory spectral data may be smoothed based on algorithms such as moving average, Savitzky-Golay filtering, Fourier transform smoothing, wavelet transform denoising, Gaussian filtering, polynomial fitting, etc.

[0133] After smoothing, the laboratory spectral data after smoothing can be divided into a training set and a validation set. The training set and the validation set can be randomly extracted from the laboratory spectral data according to a preset ratio. The preset ratio can be set comprehensively according to data capacity, resolution, and required data analysis accuracy. For example, the ratio of the training set to the validation set is 3:1.

[0134] Then, the spectral resolution of the remote sensing spectral data is obtained, and the laboratory spectral data is resampled according to the spectral resolution so that the number of bands included in the laboratory spectral data is equal to the number of bands included in the remote sensing spectral data. For example, if the remote sensing spectral data contains reflectance of 295 bands, the corresponding spectral resolution is 295, and therefore, the laboratory spectral data can be resampled based on the spectral resolution to obtain laboratory spectral data containing 295 bands.

[0135] After preprocessing the spectral data, a training set can be extracted from the spectral data divided into a training set and a validation set, that is, training spectral data can be obtained. The training spectral data is data randomly extracted from the laboratory spectral data according to a preset ratio. Then, according to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated. The tailings spectral index is the band ratio between the tailings attribute sensitive band and the tailings attribute insensitive band of the main component in the training spectral data.

[0136] The multi-objective programming algorithm is a method to solve the optimization of multiple objective functions in a given area. When solving multi-objective optimization problems, it can find a compromise solution that can approximately optimize several objectives at the same time, that is, the Pareto optimal solution set. The multi-objective programming algorithm can calculate the tailings spectral index of the training spectral data through one or more combinations of algorithms such as non-dominated sorting genetic (NSGA-II) algorithm, subgroup optimization algorithm, utility optimization model, ideal point model, constraint model, constraint model, weight method, ε-constraint method, Pareto optimization, etc.

[0137] In the process of multi-objective planning, the band ratio between the sensitive and insensitive bands in the reflectance spectrum for the iron tailings properties can be calculated based on the multi-objective planning algorithm in order to find the optimal band combination. This band combination has the largest difference in the absorption characteristics of the tailings properties, and has similar spectral characteristics for factors such as water content and particle size. For example, when using the non-dominated sorting genetic algorithm (NSGA-II) to perform multi-objective planning, the cost function and constraints used for band selection screening in the multi-objective planning algorithm are as follows:

[0138]

[0139] in, X represents the potential band combinations in the VIS-NIR range ( λ i , λ j ), λ i and λ j They are the bands that are sensitive and insensitive to the properties of iron tailings, respectively. Ω represent X The decision space represents all possible band combinations. In the embodiment of the present application, the decision space is limited to the absorption feature range of the reflectance spectrum of the tailings sample.

[0140] Based on this, F ( X ) When measuring TFe content, the objective function and equation are:

[0141]

[0142] Where RC() represents the weighting function associated with each iron tailings attribute; TMC Indicates the moisture content of tailings; TPS Indicates the tailings particle size.

[0143] F ( X ) When measuring SiO2 content, the objective function and equation are:

[0144]

[0145] Where RC() represents the weighting function associated with each iron tailings attribute; TMC Indicates the moisture content of tailings; TPS Indicates the tailings particle size.

[0146] like Figure 5 As shown, in some embodiments, in order to perform multi-objective planning, potential band combinations can be extracted from the training spectral data. And a decision space of the multi-objective planning algorithm is defined, wherein the decision space includes all band combinations within the reflectance spectrum absorption characteristic range. Then, the content objective function of the main components in the tailings sample is defined respectively, and a cost function for band selection screening is set according to the content objective function. Then, based on the cost function, the potential band combination and the decision space, the band ratio between the tailings attribute sensitive band and the tailings attribute insensitive band in the training spectral data can be calculated, and the optimal band combination can be determined according to the band ratio.

[0147] After determining the optimal band combination, characteristic bands can be extracted from the training spectral data, that is, bands where absorption characteristics appear. Then, the optimal band combination of the target substance content is obtained using the multi-objective planning algorithm. The target substance is one of the main components of the tailings sample. For example, for the iron tailings sample, the target substance is an iron-containing compound (TFe) and silicon dioxide (SiO2). The optimal band combination includes the wavelength value determined in the characteristic band according to the multi-objective planning algorithm.

[0148] Then, the correlation information between the reflectivity spectrum and the tailings properties is obtained, and the correlation information is used to characterize the mapping relationship between the wavelength and the correlation coefficient. According to the correlation information, the sensitive wavelength and the insensitive wavelength of the target substance are determined. Then, the tailings spectral index is calculated according to the sensitive wavelength and the insensitive wavelength. Among them, the tailings spectral index includes a ratio index and a three-band spectral index (TBI).

[0149] For example, Figure 6As shown in the figure, after measuring the reflectance spectra of 230 iron tailings samples collected, it can be determined based on the spectral data that the iron tailings reflectance spectra have absorption features in the 750-830nm, 900-1400nm and 2200-2400nm bands. Figure 7 As shown, by calculating the Pearson correlation between the iron tailings reflectance spectrum and the attributes, it can be seen that the iron tailings reflectance is negatively correlated with the TFe content, and the absolute value of the correlation coefficient is 0.41. The iron tailings reflectance is positively correlated with SiO2, and the absolute value of the correlation coefficient is 0.76.

[0150] Taking the characteristic band ranges of 750-830nm, 900-1400nm and 2200-2400nm of iron tailings as constraints, the multi-objective planning algorithm can be used to obtain the optimal band combination of TFe content as 827nm, 900nm and 2200nm, and the optimal band combination of SiO2 content as 776nm, 900nm and 2397nm. According to the correlation between the reflectance spectrum and the properties of iron tailings, the reflectivity of 900nm and 2200nm is relatively sensitive to TFe content, while the reflectivity of 827nm is relatively insensitive to TFe content. The reflectivity of 776nm and 900nm is relatively sensitive to SiO2 content, while the reflectivity of 2397nm is relatively insensitive to SiO2 content.

[0151] Therefore, the ratio calculation can be performed by dividing the band that is highly insensitive to the iron tailings property content by the band that is sensitive to the iron tailings property content. Figure 8 , Fig. 9 As shown, for TFe content, two ratio indices R can be calculated 827 / R 900 , R 827 / R 2200 and a three-band spectral index R 827 / (R 900 ×R 2200 ). For SiO2 content, two ratio indices R were calculated 2397 / R 776 , R 2397 / R 900 and a TBI index R 2397 / (R 776 ×R 900 ). Therefore, it can be determined that all spectral indices are positively correlated with TFe content and negatively correlated with SiO2 content.

[0152] S500: constructing a regression model according to the tailings spectral index, and using the regression model to estimate the main component content corresponding to the corrected spectral data.

[0153] After determining the tailings spectral index, a regression model can be constructed based on the tailings spectral index. The regression model can include independent variables x and the dependent variable y , where the independent variable x is the tailings spectral index, y is the content of target substance in tailings sample.

[0154] The regression model can be obtained by regression analysis calculation using the training spectral data in the training set. The regression model can be constructed based on a linear regression analysis algorithm or a nonlinear regression analysis algorithm, such as partial least squares regression (PLSR), support vector regression (SVR), and artificial neural network (ANN).

[0155] In some embodiments, after constructing a regression model based on the tailings spectral index, the constructed regression model can also be verified for accuracy, that is, after establishing a regression model between the spectral index and the content of the main components based on the training set, the regression model can be verified for accuracy using the verification set, and the regression model verification accuracy can be obtained. The regression model verification accuracy includes selecting the determination coefficient and the root mean square error. The regression model is verified for accuracy using the verification set divided from the laboratory spectral data. When performing accuracy verification, the determination coefficient (R 2 ) and root mean square error (RMSE) evaluation indicators to obtain the regression model verification accuracy.

[0156] Among them, the coefficient of determination R 2 It can be calculated according to the following formula:

[0157]

[0158] The root mean square error can be calculated as follows:

[0159]

[0160] In the formula, n is the sample size; is the predicted value of the target substance content; is the measured value of the target substance content; is the mean value of the target substance content.

[0161] For example, the calculation results of the validation accuracy of the linear regression model and nonlinear regression model constructed based on the TBI index are shown in Table 1:

[0162] Table 1. Validation accuracy results of the regression model constructed based on TBI index;

[0163]

[0164] Based on Table 1, it can be determined that the nonlinear regression model has the highest prediction accuracy for TFe and SiO2 content, R 2 They are 0.71 and 0.68 respectively, and the RMSE are 0.88% and 2.07% respectively.

[0165] like Fig.10 As shown, in some embodiments, after the regression model verification accuracy is calculated, the accuracy evaluation interval can also be obtained, and the regression model verification accuracy and the accuracy evaluation interval are compared. If the regression model verification accuracy is not within the accuracy evaluation interval, a three-band spectral index is obtained, and the regression model is reconstructed based on the three-band spectral index.

[0166] For example, R 827 / (R 900 ×R 2200 ) was in the range of 2.46-4.89, and the correlation coefficient (r) with TFe was 0.87, which was better than R 827 / R 900 (r=0.68) and R 827 / R 2200 (r=0.52) 2397 / R 776 and R 2397 / R 900 In comparison, R 2397 / (R 776 ×R 900 The TBI of the original spectra ranged from 4.58 to 11.66, with the highest correlation with SiO2 (r=0.70). TBI also had a higher correlation with TFe content than the original spectra.

[0167] If the validation accuracy of the regression model is within the accuracy evaluation interval, it means that the regression model meets the preset accuracy requirement, and therefore the regression model can be used to estimate the main component content corresponding to the corrected spectral data.

[0168] In some embodiments, in order to estimate the content of the main components of the tailings, a spatial distribution map of the main components can be drawn based on the regression model and the corrected spectral data. For example, TBI is calculated using the GF-5 satellite spectral image data corrected by DS, and then the spatial distribution map of TFe and SiO2 content is drawn using the regression model, such as Fig.11 As shown. It can be seen that the change in TBI shows that there is a clear difference in the spatial distribution between TFe and SiO2. The high TBI value area indicates that the TFe content is high and the SiO2 content is low. For all tailings ponds, the high SiO2 content is mainly distributed in the area close to the tail pipe, while the high TFe content is mainly distributed in the area far away from the tail pipe.

[0169] Since TBI is proposed based on the reflectance spectra measured on dried and sieved samples in the laboratory, in order to transfer TBI from laboratory spectral data to remote sensing spectral data, such as Fig.12 As shown, tailings samples with different particle sizes and water contents can also be prepared. All dry tailings samples can be screened using standard sieves with apertures of 0.5 mm and 0.075 mm, and based on the spectral detection and pretreatment methods provided in the above embodiments, the reflectance spectra and pretreatment are measured in a darkroom. Tailings samples with water contents of 10.94% and 20.48% are selected to compare the performance of TBI under different water contents. It can be seen that TBI is normally distributed and has no significant changes under different particle sizes and water contents, indicating that TBI can effectively reduce surface moisture and the influence of roughness caused by different particle sizes.

[0170] By applying the technical solution of this embodiment, for the tailings reflectance spectrum data, a tailings spectrum index can be constructed using a multi-objective planning algorithm to eliminate the influence of factors such as particle size and water content on the estimation of tailings properties. And using a direct standardization algorithm, the remote sensing spectrum data is corrected based on the reflectance spectrum data measured in the laboratory, which can eliminate the influence of environmental factors such as moisture in the tailings and atmospheric radiation transmission on the remote sensing reflectance spectrum data, and realize the regression model constructed based on the indoor reflectance spectrum, and transfer to the remote sensing spectrum data under field observation conditions, providing a new technical means for rapid, accurate, and large-area monitoring of tailings properties.

[0171] Further, as a specific implementation of the tailings component remote sensing spectral inversion method described in the above embodiment, the present application embodiment provides a tailings component remote sensing spectral inversion system, such as Fig.13 As shown, the system comprises:

[0172] A spectrum acquisition module is used to acquire spectrum data, wherein the spectrum data includes remote sensing spectrum data and laboratory spectrum data; the remote sensing spectrum data is data obtained by performing spectrum detection on a target area through a remote sensing device; the laboratory spectrum data is data obtained by performing spectrum detection on a tailings sample collected in the target area through a spectrometer;

[0173] A transfer function construction module, used for constructing an error transfer function between the laboratory spectral data and the remote sensing spectral data based on a direct standardization algorithm;

[0174] A correction module, used for correcting the remote sensing spectral data using the error transfer function to obtain corrected spectral data;

[0175] A planning module, for calculating, according to a multi-objective planning algorithm, a tailings spectral index of training spectral data, wherein the training spectral data is data randomly extracted from the laboratory spectral data according to a preset ratio; the tailings spectral index is a band ratio between a tailings attribute sensitive band and a tailings attribute insensitive band in the training spectral data;

[0176] The regression analysis module is used to construct a regression model according to the tailings spectral index, and use the regression model to estimate the main component content corresponding to the corrected spectral data.

[0177] It should be noted that, for other corresponding descriptions of the functional units involved in the tailings composition remote sensing spectral inversion system provided in the embodiment of the present application, reference can be made to the corresponding descriptions in the method described in the above embodiment, and will not be repeated here.

[0178] By applying the technical solution of this embodiment, the tailings component remote sensing spectral inversion system can obtain remote sensing spectral data and laboratory spectral data. And based on the direct standardization algorithm, an error transfer function between laboratory spectral data and remote sensing spectral data is constructed. The error transfer function is then used to perform correction on the remote sensing spectral data to obtain corrected spectral data. Then, according to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated. And a regression model is constructed based on the tailings spectral index, and the regression model is used to estimate the main component content corresponding to the corrected spectral data. The system can correct the remote sensing spectral data through the error transfer function, so that the regression model constructed based on the laboratory spectral data can be directly applied to the remote sensing spectral data, estimate the main component content of the tailings, and improve the accuracy of detecting the main component content of the tailings based on the remote sensing spectral data.

[0179] The embodiment of the present application also provides a computer device, which can be a personal computer, a server, a network device, etc. The computer device includes a bus, a processor, a memory and a communication interface, and can also include an input and output interface and a display device. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store location information. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps in each method embodiment are implemented.

[0180] Those skilled in the art will appreciate that the structure of the above-mentioned computer device is only a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components, or combine certain components, or have a different arrangement of components.

[0181] In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium may be non-volatile or volatile, and stores a computer program thereon. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0182] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0183] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0184] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0185] Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc.

[0186] Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0187] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0188] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0189] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A remote sensing spectral inversion method for tailings composition, characterized in that: The method comprises: Acquire spectral data, wherein the spectral data includes remote sensing spectral data and laboratory spectral data; the remote sensing spectral data is data obtained by performing spectral detection on a target area through a remote sensing device; the laboratory spectral data is data obtained by performing spectral detection on a tailings sample collected in the target area through a spectrometer; constructing an error transfer function between the laboratory spectral data and the remote sensing spectral data based on a direct normalization algorithm; Using the error transfer function to perform correction on the remote sensing spectral data to obtain corrected spectral data; According to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated, wherein the training spectral data is data randomly extracted from the laboratory spectral data according to a preset ratio; the tailings spectral index is the band ratio between the tailings attribute sensitive band and the tailings attribute insensitive band of the main component in the training spectral data; A regression model is constructed according to the tailings spectral index, and the regression model is used to estimate the main component content corresponding to the corrected spectral data.

2. The method according to claim 1, characterized in that The method further comprises: Obtaining a first sample weight, where the first sample weight is the weight weighed when the sampling container is filled with the wet tailings sample; Obtaining a second sample weight, where the second sample weight is the weight weighed when the sampling container is filled with a dry tailings sample; the dry tailings sample is obtained by drying the wet tailings sample filled in the sampling container; Obtain the weight of the sampling container; Calculating a first difference and a second difference, wherein the first difference is the difference between the weight of the first sample and the weight of the second sample; and the second difference is the difference between the weight of the second sample and the weight of the sampling container; The sample moisture content is calculated according to the first difference and the second difference. The sample moisture content is equal to the ratio of the first difference to the second difference. The sample moisture content is used to collect tailings samples with different moisture contents.

3. The method according to claim 1, characterized in that The method further comprises: Set up X-ray diffraction measurement parameters; controlling the X-ray diffraction device to emit X-rays to the ground dry tailings sample based on the X-ray diffraction measurement parameters; Acquiring diffraction data, wherein the diffraction data includes the position, diffraction intensity and diffraction peak width of the diffraction peak; The main components of the tailings sample are determined based on the diffraction data.

4. The method according to claim 1, characterized in that: Acquire spectral data, including: Set the spectrum measurement parameters; Controlling the spectrometer to emit measurement light to the dry tailings sample according to the spectral measurement parameters, wherein the spectrometer performs whiteboard calibration according to a preset calibration cycle during the measurement process; receiving at least four pieces of spectral data detected by the reflectance spectrum sensor of the spectrometer; the at least four pieces of spectral data are respectively reflectivities detected by the reflectance spectrum sensor after the dry tailings sample is rotated; An average value of at least four pieces of the spectral data is calculated to obtain laboratory spectral data of the tailings sample.

5. The method according to claim 1, characterized in that The method further comprises: performing a smoothing process on the laboratory spectral data; Dividing the smoothed laboratory spectral data into a training set and a validation set, with the ratio of the training set to the validation set being 3:1; Acquiring the spectral resolution of the remote sensing spectral data; The laboratory spectral data is resampled according to the spectral resolution to obtain the training spectral data.

6. The method according to claim 1, characterized in that According to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated, including: extracting potential band combinations from the training spectral data; defining a decision space of a multi-objective programming algorithm, wherein the decision space includes all band combinations within the reflectance spectral absorption feature range; defining a target function of the content of the main components in the tailings sample; Setting a cost function for band selection screening according to the content objective function; Based on the cost function, the potential band combination and the decision space, calculating the band ratio between the tailings property sensitive band and the tailings property insensitive band in the training spectral data; An optimal band combination is determined according to the band ratio.

7. The method according to claim 6, characterized in that According to the multi-objective planning algorithm, the tailings spectral index of the training spectral data is calculated, including: Extracting a characteristic band from the training spectrum data, wherein the characteristic band is a band where absorption characteristics appear; Using a multi-objective programming algorithm, an optimal band combination of target substance content is obtained, wherein the target substance is one of the main components of the tailings sample; the optimal band combination includes a wavelength value determined in the characteristic band according to the multi-objective programming algorithm; Obtaining correlation information between the reflectance spectrum and tailings properties, wherein the correlation information is used to characterize a mapping relationship between wavelength and correlation coefficient; Determining the sensitive wavelength and the insensitive wavelength of the target substance according to the correlation information; A tailings spectral index is calculated according to the sensitive wavelength and the insensitive wavelength, and the tailings spectral index includes a ratio index and a three-band spectral index.

8. The method according to claim 5, characterized in that A regression model is constructed according to the tailings spectral index, including: Establishing a regression model between the spectral index and the content of the main component based on the training set, wherein the regression model includes a linear regression model and a nonlinear regression model; Performing accuracy verification on the regression model using the verification set; The regression model validation accuracy is obtained, wherein the regression model validation accuracy includes selecting a determination coefficient and a root mean square error.

9. The method according to claim 8, characterized in that The method further comprises: Get the accuracy evaluation interval; If the regression model verification accuracy is within the accuracy evaluation interval, plotting a spatial distribution map of the main components based on the regression model and the corrected spectral data; If the regression model verification accuracy is not within the accuracy evaluation interval, a three-band spectral index is obtained, and the regression model is reconstructed based on the three-band spectral index.

10. A tailings composition remote sensing spectral inversion system, characterized in that: The system comprises: A spectrum acquisition module is used to acquire spectrum data, wherein the spectrum data includes remote sensing spectrum data and laboratory spectrum data; the remote sensing spectrum data is data obtained by performing spectrum detection on a target area through a remote sensing device; the laboratory spectrum data is data obtained by performing spectrum detection on a tailings sample collected in the target area through a spectrometer; A transfer function construction module, used for constructing an error transfer function between the laboratory spectral data and the remote sensing spectral data based on a direct standardization algorithm; A correction module, used for correcting the remote sensing spectral data using the error transfer function to obtain corrected spectral data; A planning module, for calculating, according to a multi-objective planning algorithm, a tailings spectral index of training spectral data, wherein the training spectral data is data randomly extracted from the laboratory spectral data according to a preset ratio; the tailings spectral index is a band ratio between a tailings attribute sensitive band and a tailings attribute insensitive band in the training spectral data; The regression analysis module is used to construct a regression model according to the tailings spectral index, and use the regression model to estimate the main component content corresponding to the corrected spectral data.

Images