Analysis method for silicate components in rock and ore based on X-ray fluorescence spectrum

Abstract

The invention relates to the technical field of component analysis, and discloses an X-ray fluorescence spectrum-based rock and ore silicate component analysis method, which comprises the following steps: S1, sample collection and pretreatment: collecting, crushing and drying an initial sample, S2, sample tabletting preparation, S3, standard sample preparation, S4, X-ray fluorescence spectrum analysis, and S5, data processing and result analysis. According to the method, a complex interference characteristic mode between elements can be automatically captured through an established deep learning model and learning of a large number of known components and fluorescence spectrum data, and in an example of correcting interference between multiple elements in rocks and mines, the model can utilize a network structure of the model and comprises an established convolutional layer, a pooling layer and a full-connection layer. Effective information is extracted from multiple groups of sample data, the interference relation between elements is accurately learned, and compared with a traditional method, the precision of analyzing the content of silicate components in rock ore is remarkably improved, so that a measurement result is closer to a real value.

Description

Technical Field

[0001] The present invention relates to the technical field of component analysis, and specifically provides a method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy. Background Art

[0002] Silicates are a large class of compounds composed of silicon, oxygen, and metal ions, occupying a core position in the field of rock and minerals. Their basic structural unit is the silicon-oxygen tetrahedron, that is, a silicon atom is located at the center, and four oxygen atoms are distributed at the vertices of the tetrahedron. These silicon-oxygen tetrahedrons are connected to each other by sharing oxygen atoms, forming diverse structures such as island-like, chain-like, layered, and framework-like. Different structures endow silicate minerals with different properties.

[0003] In the prior art, people often use the empirical coefficient method to correct the mutual interference between elements, so as to improve the accuracy in the actual analysis process. However, in the actual use process, due to the variety of elements in rock and ore samples, the interference relationships between elements are complex and not simple linear relationships. The empirical coefficient method usually corrects interference based on a linear assumption. For some complex non-linear interferences, such as self-absorption, enhancement effects generated by some elements at high concentrations, and synergistic or antagonistic effects between elements, it is difficult to accurately correct. In addition, for some unknown elements or unconsidered interference factors in the sample, the empirical coefficient method cannot effectively correct. In view of this, we propose a method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy, which solves the problem that the empirical coefficient method usually corrects interference based on a linear assumption, and it is difficult to accurately correct some complex non-linear interferences, such as self-absorption, enhancement effects generated by some elements at high concentrations, and synergistic or antagonistic effects between elements.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy, including the following steps: S1: Sample collection and pretreatment At the rock and ore sampling site, according to the geological characteristics and distribution laws, the multi-point sampling method is used to select multiple sampling points to ensure that the collected samples can represent the overall characteristics of the rock and ore in this area. At least 3 rock and ore samples are collected at each sampling point, and the distance between sampling points is not less than 5 meters. Remove the impurities on the surface of the collected rock and ore samples, and use a jaw crusher to coarsely crush them to a particle size less than 1 cm, and then finely crush them to a particle size less than 100 mesh through a ball mill. Subsequently, place the crushed samples in an oven and dry them at 105 - 110 °C for 2 - 3 hours. Then take them out and put them in a desiccator to cool to room temperature for standby; S2: Preparation of sample tablets Weigh 5 - 8 g of the dried rock and ore sample powder, add 1 - 3 g of the binder boric acid, mix them in a mixer at a speed of 200 - 300 revolutions per minute for 10 - 15 minutes. Then, put the mixed material into a tablet press and keep it under a pressure of 10 - 15 MPa for 3 - 5 minutes to make a circular thin slice with a diameter of 30 - 40 mm; S3: Preparation of standard samples Select a variety of standard rock and ore samples with compositions similar to the rock and ore samples to be measured, covering different geological origins and different lithologies. Carry out the same crushing and drying treatments on the standard rock and ore samples. The crushing particle size and drying conditions are the same as those in the pretreatment steps of the samples to be measured. Then, according to the same method as the preparation of the sample tablets to be measured, add the same proportion of the binder to prepare standard sample thin slices, and the content of the silicate component in the standard samples is known and has a gradient difference; S4: X-ray fluorescence spectrometry analysis Put the prepared sample thin slices and standard sample thin slices into the X-ray fluorescence spectrometer in turn. Before analysis, preheat the instrument for 30 - 60 minutes, set the excitation voltage to 40 - 60 kV, the excitation current to 40 - 60 mA, and the scanning time to 30 - 60 seconds. Carry out X-ray fluorescence spectrometry scanning on the samples to obtain the fluorescence intensity data of each element in the samples; S5: Data processing and result analysis Based on the fluorescence intensity data of the standard samples and the known content of the silicate component, use the least squares method to establish a calibration curve and fit the calibration curve equation. For all the obtained fluorescence intensity data, use the background correction algorithm based on spectral characteristics to carry out background correction, and at the same time carry out element interference correction based on the deep learning model. Substitute the corrected fluorescence intensity data of the samples to be measured into the calibration curve equation to calculate the content of the silicate component in the samples to be measured. Finally, carry out uncertainty evaluation on the calculated content of the silicate component, and comprehensively analyze the sources of uncertainty, including sample preparation, instrument measurement, and calibration curve fitting.

[0006] Preferably, in the sample collection and pretreatment step, a geological compass is used to determine the sampling point location, and the geographical coordinates, altitude, and occurrence information of the rock and ore at the sampling point are recorded. The geographical coordinates are accurate to six decimal places, the altitude is accurate to 0.1 m, and the occurrence information of the rock and ore includes the strike, dip, and dip angle of the rock and ore layer. The measurement accuracy of the strike and dip is ±1°, and the measurement accuracy of the dip angle is ±0.5°.

[0007] Preferably, in the sample pretreatment step, a laser particle size analyzer is used to monitor the particle size of the crushed sample in real time to ensure that the particle size is less than 100 mesh. The measurement range of the laser particle size analyzer is 0.1 - 1000 μm, and the measurement accuracy is ±1%. The particle size of the sample is measured every 5 minutes. When the particle size is less than 100 mesh in all three consecutive measurement results, the particle size of the sample is determined to be qualified.

[0008] Preferably, in the sample tablet preparation step, a release agent is applied to the inner wall of the die of the tablet press. The release agent is zinc stearate, and the application thickness is 0.05 - 0.1 mm. The release agent is evenly applied to the inner wall of the die by spraying method. After application, the die is baked in an oven at 50 - 60 °C for 10 - 15 minutes to make the release agent adhere to the surface of the die.

[0009] Preferably, in the standard sample preparation step, a complete chemical analysis is performed on the selected standard rock and ore samples to verify the accuracy of the silicate component content, and a standard sample database is established to record the detailed information of the standard samples, including the source, composition, and geological characteristics.

[0010] Preferably, a combination of multiple analysis methods is used for the complete chemical analysis, including the determination of SiO2 content by gravimetric method and the determination of Al2O3 and Fe2O3 contents by volumetric method to ensure the accuracy of the analysis results. The standard sample database is managed by a relational database management system.

[0011] Preferably, in the X-ray fluorescence spectrometry analysis step, the scanning mode of the instrument is set to continuous scanning. At the same time, during the scanning process, the ambient temperature and humidity of the instrument are monitored and controlled in real time. The temperature is controlled at 20 - 25 °C, and the relative humidity is controlled at 40% - 60%.

[0012] Preferably, in the data processing and result analysis step, a linearity test is performed on the calibration curve, and the linear relationship of the calibration curve is evaluated by calculating the correlation coefficient, and the correlation coefficient is required to be greater than 0.995.

[0013] Preferably, when calculating the correlation coefficient, at least 10 data points of standard samples are used for fitting, and a significance test is performed on the fitting result. If the correlation coefficient does not meet the requirements, the preparation and measurement processes of the standard samples are rechecked.

[0014] Preferably, in the data processing and result analysis step, the expanded uncertainty is calculated for the uncertainty evaluation result, including the expanded uncertainty at a 95% confidence probability. The calculation of the expanded uncertainty is based on the relevant standards for the evaluation and expression of measurement uncertainty, considering the influence of factors such as measurement repeatability, instrument uncertainty, and sample inhomogeneity on the measurement result.

[0015] The present invention provides a method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy. It has the following beneficial effects: 1. Through the established deep learning model, by learning a large amount of data on known components and fluorescence spectra, the present invention can automatically capture complex interference characteristic patterns between elements. For example, in the example of correcting the interference between multiple elements in rock and ore, the model can use its network structure, including convolutional layers, pooling layers, and fully connected layers, to extract effective information from multiple sets of sample data and accurately learn the interference relationship between elements. Compared with traditional methods, the accuracy of analyzing the content of silicate components in rock and ore is significantly improved, making the measurement result closer to the true value.

[0016] 2. By using intelligent algorithms such as genetic algorithms to optimize the calibration curve, the present invention automatically searches for the optimal curve fitting parameters through operations such as selection, crossover, and mutation. This method avoids the subjectivity and errors brought by manual parameter selection, enabling the calibration curve to better fit the standard sample data. For example, in the example of optimizing the linear calibration curve, after multiple generations of iteration, the individual with the highest fitness is found from the randomly generated initial individuals, and the calibration curve determined by its corresponding parameters is more reliable and universal, thereby improving the accuracy of calculating the content of silicate components in rock and ore based on the calibration curve.

[0017] 3. By calculating the expanded uncertainty for the uncertainty evaluation result, the present invention can comprehensively evaluate the measurement result. By quantifying the result fluctuation range at different percentage confidence probabilities, the inclusion interval of the measurement result at a specific confidence probability can be given, enabling users to understand the reliability of the measured value and providing a scientific basis for decision-making in fields such as geology and mineral resources, reducing risks. Description of the Drawings

[0018] Figure 1 It is a flow chart of the method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy; Figure 2 It is a schematic diagram of the S1 sample collection and pretreatment step of the present invention; Figure 3Schematic diagram of S5 data processing and result analysis of the present invention. Detailed implementation manners

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the specification of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0020] Embodiment: Please refer to the attached Figure 1 - attached Figure 3 , the embodiment of the present invention provides a method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy, including the following steps: S1: Sample collection and pretreatment At the rock and ore sampling site, according to the geological characteristics and distribution laws, multiple sampling points are selected by the multi-point sampling method to ensure that the collected samples can represent the overall characteristics of the rock and ore in this area. At least 3 rock and ore samples are collected at each sampling point, and the sampling point spacing is not less than 5 meters. Remove the surface impurities of the collected rock and ore samples, and use a jaw crusher to coarsely crush them to a particle size less than 1 cm, and then finely crush them to a particle size less than 100 mesh by a ball mill. Subsequently, place the crushed samples in an oven and dry them at 105 - 110 °C for 2 - 3 hours, and then take them out and cool them to room temperature in a desiccator for standby; S2: Sample tablet preparation Weigh 5 - 8 g of the dried rock and ore sample powder, add 1 - 3 g of the binder boric acid, mix it in a mixer at a speed of 200 - 300 revolutions per minute for 10 - 15 minutes. Then, put the mixed material into a tablet press and keep it under a pressure of 10 - 15 MPa for 3 - 5 minutes to make a circular thin slice with a diameter of 30 - 40 mm; S3: Standard sample preparation Select a variety of standard rock and ore samples with similar compositions to the rock and ore samples to be measured, covering different geological origins and different lithologies. Similarly, perform crushing and drying treatments on the standard rock and ore samples. The crushing particle size and drying conditions are the same as those in the pretreatment step of the samples to be measured. Then, according to the same method as the tablet preparation of the samples to be measured, add the same proportion of binder to prepare standard sample thin slices, and the content of silicate components in the standard samples is known and has a gradient difference; S4: X-ray fluorescence spectroscopy analysis Put the prepared sample thin slice and the standard sample thin slice into the X-ray fluorescence spectrometer in sequence. Before analysis, preheat the instrument for 30 - 60 minutes, set the excitation voltage to 40 - 60 kV, the excitation current to 40 - 60 mA, and the scanning time to 30 - 60 seconds, and perform X-ray fluorescence spectroscopy scanning on the sample to obtain the fluorescence intensity data of each element in the sample; S5: Data Processing and Result Analysis Based on the fluorescence intensity data of the standard sample and the known content of silicate components, establish a calibration curve using the least squares method and fit the calibration curve equation. For all the obtained fluorescence intensity data, perform background correction using a background correction algorithm based on spectral characteristics, and at the same time perform interference correction between elements based on a deep learning model. Substitute the corrected fluorescence intensity data of the sample to be measured into the calibration curve equation to calculate the content of silicate components in the sample to be measured. Finally, evaluate the uncertainty of the calculated content of silicate components, and comprehensively analyze the sources of uncertainty, including sample preparation, instrument measurement, and calibration curve fitting. The interference correction algorithm between elements established here is as follows: When correcting the interference between elements, the network structure includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives the X-ray fluorescence spectroscopy data and the corresponding rock and mineral component information. The convolutional layer scans the input data through a convolutional kernel to extract local features and capture the characteristic patterns of interference between elements. The pooling layer is used to reduce the data dimension and calculation amount. The fully connected layer integrates the data processed by convolution and pooling, and the output layer obtains the corrected fluorescence intensity or component content; Convolution operation: In the convolutional layer, assume the input feature map is , the convolutional kernel is , the bias is , and the calculation of the output feature map is as follows:

[0021] where are the coordinates of the output feature map; are the coordinates of the convolutional kernel; and are the sizes of the convolutional kernel; Activation function: Use the RelU (Rectified Linear Unit) function, and its formula is:

[0022] is used to introduce non-linear factors and enhance the expression ability of the network; Pooling operation: Taking max pooling as an example, assume the input feature map is , the output feature map is calculated as follows:

[0023] where and are the coordinate sets within the pooling window; Fully connected layer: Let the input vector be , the weight matrix be , and the bias be , the output vector is calculated as follows:

[0024] Loss function: The mean squared error (MSE) loss function is adopted to measure the difference between the predicted value and the true value. The formula is:

[0025] where is the number of samples; is the predicted value; is the true value.

[0026] Example: Suppose there is a scenario where it is necessary to correct the interference among three elements, silicon (Si), aluminum (Al), and iron (Fe) in rock and ore. 1000 groups of X-ray fluorescence spectroscopy data of rock and ore samples and the corresponding elemental content information are collected as the training set. Data preprocessing: Normalize the spectroscopy data so that its value is between 0 and 1, and at the same time, appropriately normalize the elemental content data.

[0027] Build a CNN model: The input layer receives spectroscopy data with a dimension of (100, 1) (assuming there are 100 features in the spectroscopy data) and elemental content data with a dimension of (3, 1) (corresponding to the three elements Si, Al, and Fe). Set two convolutional layers with convolutional kernel sizes of (3, 1) and (5, 1) respectively, both with a stride of 1. After each convolutional layer, a ReLU activation function and a max pooling layer are connected, and the pooling window size is (2, 1). Then connect two fully connected layers. The first layer has 50 neurons, and the second layer has 3 neurons (corresponding to the corrected contents of the three elements).

[0028] Train the model: Use the stochastic gradient descent (SGD) optimizer with a learning rate set to 0.001 and train for 100 epochs. During the training process, the model continuously adjusts the weights and biases to gradually reduce the loss function.

[0029] Prediction and calibration: For the spectral data of new rock and ore samples, input the trained model to obtain the predicted values of the calibrated element contents. For example, the calibrated content of Si predicted by the trained model for a certain unknown sample is 0.35 (after normalization), and the actual content is obtained by denormalization as Meanwhile, for the calibration curve, the following algorithm is proposed to optimize and calibrate it: Through operations such as selection, crossover, and mutation, optimize the parameters of the calibration curve; First, encode the parameters of the calibration curve (such as the in the and as chromosomes, then randomly generate the initial population, calculate the fitness of each individual (i.e., a set of parameters), make selections according to the fitness, so that individuals with higher fitness have a greater probability of being selected, and the selected individuals are subjected to crossover and mutation operations to generate a new population. Repeat the above process until the termination condition is met (such as reaching the maximum number of iterations or the fitness no longer changes significantly).

[0030] Formula: Function: Taking the linear calibration curve as an example, let the predicted value be , the true value be , and the fitness function is defined as:

[0031] where is the number of samples. The larger the fitness function value, the better the calibration curve corresponding to this set of parameters; Selection operation: Establish a random selection algorithm. Let the population size be , the fitness of individual be , then the probability that individual is selected is:

[0032] Crossover operation: Taking single-point crossover as an example, randomly select a crossover point, and exchange part of the genes of the two parent individuals at the crossover point to generate two offspring individuals. For example, for parent individuals and , if the crossover point is 1, then the offspring individuals and ; Mutation operation: Randomly change the genes of the individual with a certain probability. Let the mutation probability be , for individual , if a certain gene (such as ) mutates, then the new gene value , where is a random number Example: Suppose we want to optimize a linear calibration curve for analyzing the content of a certain silicate component in rock and ore. There are 20 standard rock and ore samples with known component contents and corresponding fluorescence intensities ; Initialize the population: Randomly generate 10 individuals (i.e., 10 sets of and values) as the initial population. For example, the initial individuals , etc.; Calculate the fitness: For each individual, calculate the predicted value and values, and then calculate the fitness according to the fitness function. For example, for the individual , the calculation gives:

[0033] Then the fitness is: Selection operation: According to the random selection algorithm, calculate the selection probability of each individual, and select the individuals with high fitness to enter the next generation. Suppose the selection probability of individual is 0.2, the selection probability of individual is 0.1, etc. Finally, select individuals , etc. to enter the next generation; Crossover operation: Suppose we select individuals and for crossover, with the crossover point at 1, to generate offspring individuals and ; Mutation operation: Let the mutation probability be . For the offspring individual , suppose the gene mutates, , then the mutated individual is ; Repeat iteration: Repeat the above operations of calculating fitness, selection, crossover, and mutation. After 50 generations of iteration, obtain the individual with the highest fitness, and its corresponding and values are the optimized calibration curve parameters. Suppose the finally obtained optimized parameters are , then the optimized calibration curve is .

[0034] In the sample collection and pretreatment steps, a geological compass is used to determine the sampling point location, and the geographical coordinates, altitude, and occurrence information of the rock and minerals at the sampling point are recorded. The geographical coordinates are accurate to six decimal places, the altitude is accurate to 0.1 m, and the occurrence information of the rock and minerals includes the strike, dip, and dip angle of the rock and mineral layer. The measurement accuracy of the strike and dip is ±1°, and the measurement accuracy of the dip angle is ±0.5°.

[0035] In the sample pretreatment step, a laser particle size analyzer is used to monitor the particle size of the crushed sample in real time to ensure that the particle size is less than 100 mesh. The measurement range of the laser particle size analyzer is 0.1 - 1000 μm, and the measurement accuracy is ±1%. The particle size of the sample is measured every 5 minutes. When the particle size is less than 100 mesh in three consecutive measurement results, it is determined that the particle size of the sample is qualified.

[0036] In the sample tablet preparation step, a release agent is applied to the inner wall of the die of the tablet press. The release agent is zinc stearate, and the coating thickness is 0.05 - 0.1 mm. The spray method is used to evenly apply it to the inner wall of the die. After application, the die is baked in an oven at 50 - 60 °C for 10 - 15 minutes to make the release agent adhere to the surface of the die.

[0037] In the standard sample preparation step, a chemical total analysis is performed on the selected standard rock and mineral samples to verify the accuracy of the content of their silicate components, and a standard sample database is established to record the detailed information of the standard samples, including the source, composition, and geological characteristics.

[0038] The chemical total analysis combines multiple analysis methods, including the determination of SiO2 content by the gravimetric method and the determination of Al2O3 and Fe2O3 contents by the volumetric method to ensure the accuracy of the analysis results. The standard sample database is managed by a relational database management system.

[0039] In the X-ray fluorescence spectrometry analysis step, the scanning mode of the instrument is set to continuous scanning. At the same time, during the scanning process, the environmental temperature and humidity of the instrument are monitored and controlled in real time. The temperature is controlled at 20 - 25 °C, and the relative humidity is controlled at 40% - 60%.

[0040] In the data processing and result analysis step, a linearity test is performed on the calibration curve, and the linear relationship of the calibration curve is evaluated by calculating the correlation coefficient. The correlation coefficient is required to be greater than 0.995.

[0041] When calculating the correlation coefficient, at least 10 data points of the standard samples are used for fitting, and a significance test is performed on the fitting result. If the correlation coefficient does not meet the requirements, the preparation and measurement processes of the standard samples are rechecked.

[0042] In the data processing and result analysis steps, the expanded uncertainty is calculated for the uncertainty evaluation result, including the expanded uncertainty at a 95% confidence probability. The calculation of the expanded uncertainty is based on the relevant standards for the evaluation and expression of measurement uncertainty, considering the impacts of factors such as measurement repeatability, instrument uncertainty, and sample inhomogeneity on the measurement result. The algorithm established here is as follows:

[0043] Where is the expanded uncertainty; is the coverage factor (for a 95% confidence probability, when the distribution of the measured quantity is approximately normal, ); is the combined standard uncertainty, and the combined standard uncertainty is calculated according to the law of propagation of uncertainty and is obtained by synthesizing each uncertainty component :

[0044] Where is the uncertainty propagation coefficient (related to the measurement model); is the th uncertainty component (such as the measurement repeatability uncertainty , ; For example, if the measurement repeatability uncertainty , the instrument uncertainty , and the sample inhomogeneity uncertainty are known, and assuming the uncertainty propagation coefficient (example case); Then the combined standard uncertainty:

[0045] For a 95% confidence probability, take , then the expanded uncertainty:

[0046] Example: Suppose the measurement result of the content of a certain silicate component in a rock and ore sample is , and the expanded uncertainty is obtained through the above calculation. Then the measurement result can be expressed as (at a 95% confidence probability), that is, the content of this silicate component in the rock and ore sample is between and The probability is 95%.

[0047] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy, characterized in that, It includes the following steps: S1: Sample collection and pretreatment At the rock and ore sampling site, according to the geological characteristics and distribution rules, the multi-point sampling method is used to select multiple sampling points to ensure that the collected samples can represent the overall characteristics of the rock and ore in this area. At least 3 rock and ore samples are collected at each sampling point, and the distance between sampling points is not less than 5 meters. Remove the surface impurities of the collected rock and ore samples, and use a jaw crusher to coarsely crush them to a particle size less than 1 cm, and then finely crush them to a particle size less than 100 mesh through a ball mill. Subsequently, place the crushed samples in an oven and dry them at 105 - 110 °C for 2 - 3 hours, and then take them out and cool them to room temperature in a desiccator for standby; S2: Preparation of sample tablets Weigh 5 - 8 g of the dried rock and ore sample powder, add 1 - 3 g of the binder boric acid, mix them in a mixer at a speed of 200 - 300 revolutions per minute for 10 - 15 minutes. Then, put the mixed material into a tablet press and keep it under a pressure of 10 - 15 MPa for 3 - 5 minutes to make a circular thin slice with a diameter of 30 - 40 mm; S3: Preparation of standard samples Select a variety of standard rock and ore samples with similar compositions to the rock and ore samples to be measured, covering different geological origins and different lithologies. Similarly, carry out crushing and drying treatments on the standard rock and ore samples. The crushing particle size and drying conditions are the same as those in the pretreatment step of the samples to be measured. Then, according to the same method as the preparation of the sample tablets to be measured, add the same proportion of binder to prepare standard sample thin slices, and the content of silicate components in the standard samples is known and has a gradient difference; S4: X-ray fluorescence spectrometry analysis Put the prepared sample thin slices and standard sample thin slices into the X-ray fluorescence spectrometer in turn. Before analysis, preheat the instrument for 30 - 60 minutes, set the excitation voltage to 40 - 60 kV, the excitation current to 40 - 60 mA, and the scanning time to 30 - 60 seconds, and perform X-ray fluorescence spectrometry scanning on the samples to obtain the fluorescence intensity data of each element in the samples; S5: Data processing and result analysis Based on the fluorescence intensity data of the standard samples and the known content of silicate components, use the least squares method to establish a calibration curve and fit the calibration curve equation. For all the obtained fluorescence intensity data, perform background correction using a background correction algorithm based on spectral characteristics, and at the same time perform interference correction between elements based on a deep learning model. Substitute the corrected fluorescence intensity data of the samples to be measured into the calibration curve equation to calculate the content of silicate components in the samples to be measured. Finally, conduct uncertainty evaluation on the calculated content of silicate components, and comprehensively analyze the sources of uncertainty, including sample preparation, instrument measurement, and calibration curve fitting.

2. The method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the sample collection and pretreatment step, a geological compass is used to determine the positions of the sampling points, and the geographical coordinates, altitude, and occurrence information of the rock and ore are recorded. The geographical coordinates are accurate to six decimal places, the altitude is accurate to 0.1 m, and the occurrence information of the rock and ore includes the strike, dip, and dip angle of the rock and ore layer. The measurement accuracy of the strike and dip is ±1°, and the measurement accuracy of the dip angle is ±0.5°.

3. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the sample pretreatment step, a laser particle size analyzer is used to monitor the particle size of the crushed sample in real time. It is checked that the particle size is less than 100 mesh. The measurement range of the used laser particle size analyzer is 0.1 - 1000 μm, and the measurement accuracy is ±1%. The particle size of the sample is measured every 5 minutes. When the particle size is shown to be less than 100 mesh in three consecutive measurement results, it is determined that the particle size of the sample is qualified.

4. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the sample tablet pressing preparation step, a release agent is applied to the inner wall of the die of the tablet press. The release agent is zinc stearate, and the coating thickness is 0.05 - 0.1 mm. The spray method is used to evenly apply it on the inner wall of the die. After application, the die is baked in an oven at 50 - 60 °C for 10 - 15 minutes to make the release agent adhere to the surface of the die.

5. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the standard sample preparation step, a chemical total analysis is performed on the selected standard rock and ore samples to verify the accuracy of the content of their silicate components, and a standard sample database is established to record the detailed information of the standard samples, including the source, composition, and geological characteristics.

6. The method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 5, wherein The chemical total analysis combines multiple analysis methods, including the gravimetric method for determining the SiO2 content, the volumetric method for determining the Al2O3 and Fe2O3 contents. The standard sample database is managed using a relational database management system.

7. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the X-ray fluorescence spectrometry analysis step, the scanning mode of the instrument is set to continuous scanning. At the same time, during the scanning process, the ambient temperature and humidity of the instrument are monitored and controlled in real time. The temperature is controlled at 20 - 25 °C, and the relative humidity is controlled at 40% - 60%.

8. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the data processing and result analysis step, a linearity test is performed on the calibration curve, and the linear relationship of the calibration curve is evaluated by calculating the correlation coefficient. It is required that the correlation coefficient is greater than 0.

995.

9. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 8, characterized in that, When calculating the correlation coefficient, at least 10 data points of the standard samples are used for fitting, and a significance test is performed on the fitting result. If the correlation coefficient does not meet the requirements, the preparation and measurement processes of the standard samples are rechecked.

10. A method for analyzing silicate components in rock and ore based on X-ray fluorescence spectroscopy according to claim 1, characterized in that, In the data processing and result analysis step, an expanded uncertainty calculation is performed on the uncertainty evaluation result, including the expanded uncertainty under a 95% confidence probability. The calculation of the expanded uncertainty is based on the standard for the evaluation and expression of measurement uncertainty, considering the influence of factors such as measurement repeatability, instrument uncertainty, and sample inhomogeneity on the measurement result.

Images