Lithology complex area lithium geochemical anomaly identification method and system

Abstract

The present application is suitable for the field of mineral exploration, and provides a lithium geochemical anomaly identification method and system for a lithology complex area, which comprises the following steps: obtaining geochemical data of the area, and preprocessing the geochemical data of the area; determining PLSR independent variable indexes according to the preprocessed geochemical data; constructing a PLSR regression model according to the PLSR independent variable indexes; determining lithium geochemical background values of each sample point in the area according to the PLSR regression model; and identifying lithium geochemical anomalies in the area according to the predicted upper limit of the lithium geochemical background values. The present application uses PLSR to construct a regression model between lithium and lithology indicating elements, and then can determine the lithium geochemical background values of each sample point, effectively solves the multicollinearity problem in lithium multiple linear regression analysis, improves the calculation accuracy of the lithium geochemical background values of each sample point in the lithology complex area, and lays a solid foundation for lithium geochemical anomaly identification.

Description

Technical Field

[0001] This invention belongs to the field of mineral exploration, and in particular relates to a method and system for identifying lithium geochemical anomalies in lithologically complex areas. Background Technology

[0002] Lithium (Li) is hailed as a "green energy metal" and "white oil," with wide applications in energy storage, metallurgy, and chemicals. In recent years, with the rapid development of the new energy vehicle industry, the demand for lithium resources both domestically and internationally has exploded, gradually increasing the need for lithium mineral resource exploration. In mineral resource exploration, geochemical anomalies are a crucial factor in determining whether further exploration can proceed, making accurate identification of geochemical anomalies of great significance. However, in the identification of lithium geochemical anomalies in lithologically complex areas, variations in elemental background are often a concern.

[0003] Scholars have proposed various solutions to the problem of varying element backgrounds, including the probabilistic lattice method, the fractal method, and unsupervised machine learning algorithms. k Algorithms such as -means clustering and EM clustering have played an important role in addressing elemental background variation, but they all share a common limitation: they cannot obtain the lithium geochemical background value for each sample point. These methods first classify regional geochemical data or samples, and then calculate the lithium background value for each class of samples separately, also known as the local background value. Although this type of method can reduce the influence of lithium background variation, the method of calculating local background values ​​by clustering samples is somewhat crude and insufficient to reflect the background variation of lithium across the entire region.

[0004] Multiple linear regression analysis is a highly effective method for addressing elemental background variation problems, allowing the determination of lithium geochemical background values ​​for each sample point. This method first explores the linear relationship between lithium and lithological indicator elements (such as Si and Al), then establishes a linear regression equation to estimate the lithium background value, effectively eliminating the influence of lithium background variation. However, because geochemical data are typically compositional data with a closure effect, significant correlations often exist between elements. These correlations can lead to multicollinearity in regression analysis, significantly affecting the accuracy of regression coefficients. Multicollinearity limits the application of this method in lithium anomaly identification.

[0005] In summary, determining the geochemical background of elements is a prerequisite for identifying geochemical anomalies. Existing technologies such as probabilistic lattice paper methods, fractal methods, and unsupervised machine learning algorithms can estimate the local background value of lithium, but cannot obtain the background value for each sample point. In addition, multiple linear regression analysis can obtain the lithium geochemical background value for each sample point, but it is significantly affected by multicollinearity, resulting in inaccurate estimation of lithium background values. SUMMARY

[0006] The present application aims to provide a method for identifying lithium geochemical anomalies in complex lithology areas, and aims to solve the above technical problems.

[0007] The present application is implemented by a method for identifying lithium geochemical anomalies in complex lithology areas, comprising the following steps:

[0008] Obtain geochemical data of the area and preprocess the geochemical data of the area;

[0009] Determine PLSR independent variable indicators according to the preprocessed geochemical data;

[0010] Construct a PLSR regression model according to the PLSR independent variable indicators;

[0011] Determine the lithium geochemical background value of each sample point in the area according to the PLSR regression model;

[0012] Identify the lithium geochemical anomalies in the area according to the predicted upper limit of the lithium geochemical background value.

[0013] Further, the step of preprocessing the geochemical data specifically comprises:

[0014] The central logarithmic ratio transformation method is used to process the geochemical data of the area, and then the 3σ method is used to iteratively remove outliers in the data.

[0015] Further, the step of determining PLSR independent variable indicators according to the preprocessed geochemical data specifically comprises:

[0016] Screen out lithology indicator elements such as SiO2, Al2O3, Fe2O3, MgO, CaO, Na2O, K2O, Cr, Zr, Sr, etc. from the preprocessed geochemical data, which can reflect lithology and are not affected by mineralization, to form an initial independent variable set.

[0017] Calculate the regression coefficient of each independent variable in the initial independent variable set, iteratively remove lithology indicator elements with regression coefficients less than a preset threshold, and determine PLSR independent variable indicators.

[0018] Further, the step of constructing a PLSR regression model according to the PLSR independent variable indicators specifically comprises:

[0019] Extracting component variables: according to the PLSR independent variable index, let X be an n x p independent variable matrix, Y be an n x 1 dependent variable vector, and Y be a lithium content data matrix; find the weight vector w1 of the independent variable X in the high-dimensional space, construct the first component variable t1=Xw1, and solve the weight vector w1 by maximizing the covariance max cov(t1,Y);

[0020] Establishing a regression model: , wherein c1 is the regression coefficient vector of Y, and E1 is the residual matrix of Y; regress the independent variable X: X t 1 q 1 T + F 1, wherein q1 is the regression coefficient vector of X, and F1 is the residual matrix of X;

[0021] Iterative extraction of multiple component variables: repeat the above steps for the residual matrices E1 and F1 to extract the next pair of component variables t2 and w2, and so on, until a sufficient number of component variables are extracted; the PLSR regression model of lithium is represented as: L i =XWC T +E; wherein L i is the lithium geochemical background value, W is a matrix composed of weight vectors , C is a matrix composed of regression coefficient vectors , and E is the final residual matrix;

[0022] Determining the optimal number of component variables: construct the determination coefficient R 2 and the prediction determination coefficient R p 2 Two evaluation indexes, when R 2 and / or R p 2 reach the maximum, and R 2 and R p 2 are closest, the optimal number of component variables can be obtained; wherein R p 2 can be determined by leave-one-out cross-validation method;

[0023] ;

[0024] ;

[0025] According to the optimal number of component variables, determine the PLSR regression model of the final lithium geochemical background value.

[0026] Further, according to the prediction upper limit of the lithium geochemical background value, the steps of identifying the lithium geochemical anomaly of the region include: ​

[0027] According to the PLSR regression model of lithium, the predicted upper limit of the lithium geochemical background value of each sample point is calculated;

[0028] The predicted upper limit of the lithium geochemical background value of each sample point is taken as the background upper limit of the corresponding sample point;

[0029] The difference between the actual observation value and the background upper limit is calculated to obtain a predicted residual value;

[0030] According to the predicted residual value, the lithium geochemical anomaly in the region is determined.

[0031] Another object of the present application is to provide a lithium geochemical anomaly identification system for a lithology complex region, which is used to implement the above-mentioned lithium geochemical anomaly identification method, and comprises:

[0032] A data preprocessing module is configured to obtain geochemical data of the region and pre-process the geochemical data of the region;

[0033] A independent variable determination module is configured to determine PLSR independent variable indexes according to the pre-processed geochemical data;

[0034] A regression model construction module is configured to construct a PLSR regression model according to the PLSR independent variable indexes;

[0035] A background value determination module is configured to determine lithium geochemical background values of each sample point in the region according to the PLSR regression model;

[0036] An anomaly identification module is configured to identify lithium geochemical anomalies in the region according to the predicted upper limit of the lithium geochemical background value.

[0037] The lithium geochemical anomaly identification method for a lithology complex region provided by the present application uses PLSR to construct a regression model between lithium and lithology indicator elements, and then can determine the lithium geochemical background value of each sample point, effectively solves the problem of multicollinearity in lithium multivariate linear regression analysis, improves the calculation accuracy of the lithium geochemical background value of each sample point in a lithology complex region, and lays a solid foundation for lithium geochemical anomaly identification. In addition, the present application can eliminate meaningless or false lithium geochemical anomalies in a high background region, reduce unnecessary subsequent anomaly verification investment, and reduce exploration risk; and can identify low and weak lithium geochemical anomalies in a low background region, and improve the success rate of exploration. Under the current situation that surface mines and shallow mines are decreasing, and the focus of mineral resources exploration is constantly shifting to the search for deep mines and concealed mines, low and weak geochemical anomalies are often more meaningful. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 A flowchart of the lithium geochemical anomaly identification method for a lithology complex region provided by the present application is shown.

[0039] Figure 2 is a graph of the relationship between the number of component variables and the determination coefficient, the predicted determination coefficient.

[0040] Figure 3 is a spatial variation graph of the lithium geochemical background value.

[0041] Figure 4 is a lithium geochemical anomaly map; in the figure, a is the lithium geochemical anomaly identification result based on PLSR, and b is the lithium geochemical anomaly identification result based on the [mean + 2 standard deviation] method.

[0042] Figure 5 is a structural schematic diagram of a lithium geochemical anomaly identification system for a lithology complex region provided by an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present application clearer and more comprehensible, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0044] As shown in Figure 1 , in one embodiment of the present application, a lithology complex region lithium geochemical anomaly identification method based on partial least squares regression (PLSR) is provided, which specifically includes the following steps:

[0045] S100, acquiring geochemical data of a region and pre-processing the geochemical data of the region;

[0046] S200, determining PLSR independent variable indexes according to the pre-processed geochemical data;

[0047] S300, constructing a PLSR regression model according to the PLSR independent variable indexes;

[0048] S400, determining a lithium geochemical background value of each sample point of the region according to the PLSR regression model;

[0049] S500, identifying a lithium geochemical anomaly of the region according to a predicted upper limit of the lithium geochemical background value.

[0050] In the embodiment of the present application, a regression model between lithium and lithology indicator elements is established by using PLSR, the lithium geochemical background value of each sample point is calculated, and then the lithium geochemical anomaly can be circled, the multiple collinearity problem in lithium multiple linear regression analysis is effectively solved, the calculation accuracy of the lithium geochemical background value of each sample point in the lithology complex region is improved, and a solid foundation is laid for lithium geochemical anomaly identification.

[0051] In a preferred embodiment of the present application, the step of preprocessing geochemical data specifically comprises:

[0052] The geochemical data of the region is processed by using the center log ratio transformation method, and then the 3σ method is used to iteratively eliminate outliers in the data.

[0053] In practical application, in order to eliminate the influence of the "closure effect", the geochemical data of the region is processed by using the center log ratio transformation method in the embodiment of the present application, which not only retains all element information, but also has a certain robustness to outliers. The specific calculation formula for a single sample is as follows:

[0054] ;

[0055] Wherein, clr(x) is the data after center log ratio transformation; x is the element content; x1, x2...xp are the element contents of the 1st, 2nd...pth variables; p respectively; is the geometric mean of the sample; then, the 3σ (standard deviation) technique is used to eliminate outliers, and the calculation process includes the following two steps: first, calculate the mean and standard deviation of the original data; then, eliminate the values outside the interval [mean ± 3 standard deviation]; repeat the above operation until all data is within the above interval.

[0056] In a preferred embodiment of the present application, according to the preprocessed geochemical data, the step of determining the PLSR independent variable index specifically comprises:

[0057] S210, screening out lithology indicating elements reflecting lithology and not affected by mineralization from the preprocessed geochemical data to form an initial independent variable set;

[0058] S220, calculating the regression coefficient of each independent variable in the initial independent variable set, iteratively eliminating lithology indicating elements with regression coefficients less than a preset threshold, and determining the PLSR independent variable index.

[0059] In practical application, to establish a PLSR regression model of lithium, the independent variables of the regression model need to be determined first. Specifically, those lithology indicating elements that can reflect lithology, have high stability, and are not affected by mineralization are screened out, such as Cr, La, Ni, Sr, Th, Ti, U, V, Y, Zr, Al2O3, CaO, Fe2O3, K2O, MgO, Na2O and SiO2, etc.; then, elements with small contribution to modeling are iteratively eliminated through the regression coefficients of the independent variables; wherein, the larger the regression coefficient of an element, the greater its contribution to modeling, and vice versa.

[0060] In a preferred embodiment of the present application, according to the PLSR independent variable index, the step of constructing a PLSR regression model specifically comprises:

[0061] S310, extracting component variables: according to the PLSR independent variable index, X is an nxp independent variable matrix, Y is an nxl dependent variable vector, Y is specifically a lithium content data matrix; a weight vector w1 of the independent variable X is found in a high-dimensional space, a first component variable t1=Xw1 is constructed, and the weight vector w1 is solved by maximizing the covariance max cov(t1,Y);

[0062] S320, establishing a regression model: , wherein c1 is the regression coefficient vector of Y, and E1 is the residual matrix of Y; the independent variable X is regressed: X = t 1 q 1 T + F 1, wherein q1 is the regression coefficient vector of X, and F1 is the residual matrix of X;

[0063] S330, iteratively extracting multiple component variables: repeating the above steps for the residual matrix E1 and F1 to extract the next pair of component variables t2 and w2, and so on, until a sufficient number of component variables is extracted; the PLSR regression model of lithium is represented as: L i =XWC T +E; wherein L i is the lithium geochemical background value, W is a matrix composed of weight vectors , C is a matrix composed of regression coefficient vectors , and E is the final residual matrix;

[0064] S340, determining the optimal number of component variables: constructing the determination coefficient R 2 and the prediction determination coefficient R p 2 two evaluation indexes, when R 2 and / or R p 2 reach the maximum, and R 2 and R p 2 are closest, the optimal number of component variables can be obtained; wherein R p 2 can be determined by leave-one-out cross-validation method;

[0065] ;

[0066] ;

[0067] , the prediction residual sum of squares is equal to , the residual sum of squares is equal to , and the total sum of squares of actual observations is equal to , is the actual observation value of the first sample, is the predicted value of the first sample when the first sample is removed as a test sample, is the average value of all actual observation values, is the total number of samples.

[0068] It should be noted that the closer the value of the prediction determination coefficient R p 2 to 1, the stronger the prediction ability of the PLSR regression model; the closer the value of the determination coefficient R 2 to 1, the stronger the explanation ability of the PLSR regression model for the dependent variable.

[0069] S350, determining the PLSR regression model of the final lithium geochemical background value according to the optimal number of component variables.

[0070] In one preferred embodiment of the present application, the step of identifying the lithium geochemical anomaly of the region according to the predicted upper limit of the lithium geochemical background value specifically comprises:

[0071] S510, calculating the predicted upper limit of the lithium geochemical background value of each sample point according to the PLSR regression model of lithium, which represents the reasonable maximum value that the prediction target can reach under a specified confidence level (usually 95%). The predicted upper limit = predicted value + (t x predicted standard error), t is the critical value of the t distribution with a degree of freedom of (n-p-1), and the predicted standard error includes the overall error of the model and the uncertainty of the new observation value.

[0072] S520, taking the predicted upper limit of the lithium geochemical background value of each sample point as the background upper limit of the corresponding sample point;

[0073] S530, calculating the difference between the actual observation value and the background upper limit to obtain the predicted residual value;

[0074] S540, determining the lithium geochemical anomaly in the region according to the predicted residual value.

[0075] ​​​​​​In practical applications, the lithium geochemical background value of each sample point can be determined based on the PLSR regression model of the aforementioned lithium geochemical background value. Considering the uncertainty of the parameter estimation of the PLSR regression model and the random variability of the data itself, the upper limit of the predicted background value can be used to identify anomalies: the upper limit of the predicted background value of each sample point fitted by PLSR is taken as the upper limit of the background of the corresponding sample point, and the difference between the actual observed value and the upper limit of the background is calculated to obtain the predicted residual value. Samples with a predicted residual value greater than 0 can be regarded as anomalous samples, and thus the lithium geochemical anomaly in the region can be determined.

[0076] Taking the identification of lithium geochemical anomalies in river sediments in northeastern Hunan as an example, this paper details the application process and results of the present invention.

[0077] I. Data Preprocessing: Data preprocessing is crucial for PLSR modeling. Before modeling, central logarithmic ratio transformation was performed on 1:200,000 stream sediment geochemical data of northeastern Hunan. Then, the 3σ technique was used to iteratively remove outliers from the data to ensure data quality.

[0078] II. Construction of the PLSR regression model for lithium:

[0079] 1. Determining Independent Variables: The study area contains granite, monzogranite, granodiorite, sandstone, siltstone, mudstone, slate, and other complex lithologies. Therefore, a sufficient number of elements need to be selected as independent variables to better reflect the lithological characteristics of the study area. Based on the above analysis and the correlation coefficient analysis results of lithium with other elements, lithological indicator elements such as Cr, La, Ni, Sr, Th, Ti, U, V, Y, Zr, Al2O3, CaO, Fe2O3, K2O, MgO, Na2O, and SiO2 were initially selected as independent variables. Then, through multiple iterations, the contribution (regression coefficient) of each independent variable to the model was calculated, and elements with small regression coefficients were eliminated, as they had little effect on the model performance but could lead to overfitting. After multiple iterations to screen the independent variables, the independent variable indicators for the lithium PLSR regression model were finally determined to be Sr, Ti, V, Y, Al2O3, CaO, Fe2O3, K2O, and MgO.

[0080] 2. Model Performance Evaluation: After determining the independent variable indicators, a PLSR regression model for lithium was established. The coefficient of determination R was constructed using the method described above. 2 With the predictive coefficient of determination R p 2 Two evaluation indicators, such as Figure 2 As shown, the R-squared values ​​of the model are displayed for different numbers of component variables. 2 (Fitting) and R p 2The relationship between the number of components and the R 2 and R p 2 increases first and then decreases; when the number of components reaches 6, R p 2 reaches the maximum value. Although R 2 has not reached the maximum value at this time, the difference between the two indicators is the smallest. Therefore, according to the optimal model selection criteria, the PLSR regression model is considered to be the optimal at this stage.

[0081] 3. Determination of the PLSR regression model of lithium: After determining that the optimal number of components is 6, the PLSR regression equation of the lithium geochemical background value is finally established: L i (background value) = 3.83750-0.17707xSr-0.35090xTi-0.17547xV-0.32998xY+0.49467xAl203-0.08215xCaO-0.30069xFe203+0.10333xK20+0.25553xMgO.

[0082] III. Lithium anomaly identification based on the PLSR regression model: According to the above PLSR regression equation, the lithium geochemical background value of each sample point is calculated, and the lithium geochemical background values of different samples are significantly different (as shown in Figure 3 ). These differences are closely related to the lithology background corresponding to the sample points. Specifically, the samples in the northeast and southwest of the study area generally have high lithium geochemical background values, and the lithology in these two regions is mainly granitic rock. In the multi-stage differentiation process of magma, lithium will gradually enrich, thereby forming a high background value region. If a unified background value is used to delineate anomalies in the whole region, these high background regions may be misjudged as anomaly areas, but in fact their high background values are not caused by mineralization but are a natural result of the geochemical properties of lithium. On the contrary, the metamorphic sandstone, slate and sandstone areas in the central part of the study area are low background value regions. If a unified background value is used to delineate anomalies, it is likely to overlook the low and weak anomalies in these regions that may reflect mineralization.

[0083] The predicted upper limit of the lithium geochemical background value of each sample point is taken as the background upper limit of the corresponding region, and the difference between the actual observation value and the background upper limit, i.e. the predicted residual value, is calculated. The sample with a predicted residual value greater than 0 is an anomaly sample. Through calculation, 115 lithium anomaly samples are obtained. The predicted residual values of these anomaly samples are arranged in ascending order, and 0 is taken as the anomaly lower limit. The predicted residual values corresponding to 50% and 75% of the cumulative number of anomaly samples are taken as the division standards of medium and high anomalies, and the lithium geochemical anomaly map is drawn, as shown in Figure 4 a of the figure.

[0084] Four, abnormal recognition effect evaluation: in order to evaluate the lithium abnormal recognition effect of PLSR regression model, the traditional [mean + 2 standard deviation] method is used to identify the lithium anomaly in the study area, as shown in the b of Figure 4 [mean + 2 standard deviation] method adopts 3σ technology to remove outliers, and then uses the formula: robust mean + 2 robust standard deviation to calculate the upper limit value of the unified background of the whole area, which is used to identify lithium geochemical anomaly. By comparing and analyzing the lithium geochemical anomalies identified by PLSR regression model method and traditional method, it can be found that the PLSR regression model identification method provided by the embodiment of the present application significantly improves the identification accuracy of lithium anomaly in the study area, mainly in the following two aspects:

[0085] (1) enhance the correspondence between the anomaly and the known lithium deposit (point): compared with the traditional method, the anomaly identified by PLSR regression model has stronger spatial correlation with the known lithium deposit (point). For example, A1 and A3 anomalies circled by PLSR correspond to two lithium deposits, while the anomaly circled by the traditional method cannot identify these deposits. The reason may be that the two deposits are located in the low background area, and the anomalies in these areas may be covered by the stronger anomalies in other areas.

[0086] (2) effectively reduce false anomaly: PLSR regression model significantly reduces the meaningless anomaly in B1 and B2 areas, which are located in the granite intrusion. Due to the unique geochemical properties of lithium, the background value of lithium naturally increases in these areas. The traditional method identifies the granite intrusion in the southwest of the study area as lithium anomaly; however, there is no proven deposit in these areas. PLSR regression model successfully narrows down the anomaly range and focuses on specific areas within the granite and areas in contact with surrounding rocks. Reducing such false anomalies in high background areas is crucial for subsequent exploration work deployment.

[0087] As shown in Figure 5 , in another embodiment of the present application, a lithium geochemical anomaly identification system for lithology complex area is also provided, which is used to realize the above-mentioned lithium geochemical anomaly identification method, and the system comprises:

[0088] A data preprocessing module 10 is used to obtain geochemical data of the region and preprocess the geochemical data of the region;

[0089] A dependent variable determination module 20 is used to determine PLSR dependent variable index according to the preprocessed geochemical data;

[0090] A regression model construction module 30 is used to construct PLSR regression model according to PLSR dependent variable index;

[0091] The background value determination module 40 is configured to determine the lithium geochemical background value of each sample point in the region according to the PLSR regression model.

[0092] The anomaly identification module 50 is configured to identify the lithium geochemical anomaly of the region according to the predicted upper limit of the lithium geochemical background value.

[0093] It should be noted that the above modules can be implemented in the form of a computer program, the computer program can run on a computer device, and the computer program composed of the modules can be stored in the memory of the computer device to enable the processor to execute the steps of the above method.

[0094] It should be understood that, although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise stated herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these sub-steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0095] It can be understood by those skilled in the art that all or part of the above-mentioned embodiment methods can be completed by a computer program to instruct related hardware. The above-mentioned program can be stored in a non-volatile computer readable storage medium, and the program can include the above-mentioned flow of each method embodiment when executed. Any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include non-volatile and / or volatile memory.

[0096] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it cannot be understood as a limitation on the patent scope of the present application. It should be noted that, for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A method for identifying a lithium geochemical anomaly in a lithologically complex area, characterized in that, The method comprises the following steps: obtaining geochemical data of a region and preprocessing the geochemical data of the region; determining PLSR independent variable indexes according to the preprocessed geochemical data; constructing a PLSR regression model according to the PLSR independent variable indexes; determining lithium geochemical background values of each sample point in the region according to the PLSR regression model; identifying lithium geochemical anomalies in the region according to a predicted upper limit of the lithium geochemical background values; The step of identifying lithium geochemical anomalies in the region according to a predicted upper limit of the lithium geochemical background values specifically comprises: calculating a predicted upper limit of the lithium geochemical background values of each sample point according to the PLSR regression model of lithium; taking the predicted upper limit of the lithium geochemical background values of each sample point as the background upper limit of the corresponding sample point; the predicted upper limit represents a reasonable maximum value that the target may reach under a specified confidence level; the predicted upper limit = predicted value + (t x predicted standard error), t is a critical value of a t-distribution with a degree of freedom of (n-p-1), and the predicted standard error includes the overall error of the model and the uncertainty of a new observation value; calculating a difference between an actual observation value and the background upper limit to obtain a predicted residual value; determining lithium geochemical anomalies in the region according to the predicted residual value.

2. The method for identifying lithium geochemical anomalies in lithologically complex areas according to claim 1, characterized in that, The step of preprocessing the geochemical data specifically comprises: processing the geochemical data of the region by using a central logarithmic ratio transformation method, and then iteratively removing abnormal values in the data by using a 3σ method.

3. The method for identifying a lithium geochemical anomaly in a lithologically complex area according to claim 1, characterized in that, The step of determining PLSR independent variable indexes according to the preprocessed geochemical data specifically comprises: screening out lithology indicator elements that can reflect lithology and are not affected by mineralization from the preprocessed geochemical data to form an initial independent variable set; calculating a regression coefficient of each independent variable in the initial independent variable set, iteratively removing lithology indicator elements with a regression coefficient less than a preset threshold, and determining PLSR independent variable indexes.

4. The method for identifying a lithium geochemical anomaly in a lithologically complex area according to claim 1, characterized in that, The step of constructing a PLSR regression model according to the PLSR independent variable indexes specifically comprises: extracting component variables: taking X as an n x p independent variable matrix and Y as an n x 1 dependent variable vector, Y is specifically a lithium content data matrix; finding a weight vector w1 of the independent variable X in a high-dimensional space to construct a first component variable t1 = Xw1, and solving the weight vector w1 by maximizing the covariance max cov(t1, Y); A regression model is established: wherein c1 is a regression coefficient vector of Y, and E1 is a residual matrix of Y; regression is performed on the independent variable X: X = t 1 q 1 T + F 1, wherein q1 is a regression coefficient vector of X, and F1 is a residual matrix of X; Extracting multiple component variables iteratively: repeat the above steps on the residual matrix E1, F1 to extract the next pair of component variables t2 and w2, and so on, until the number of component variables is sufficient; the PLSR regression model of lithium is represented as: L i = XWC T + E; wherein, L i is the geochemical background value of lithium, W is a matrix composed of weight vectors , C is a matrix composed of regression coefficient vectors , and E is the final residual matrix; Determine the optimal number of components: build the coefficient of determination R 2 and the predicted coefficient of determination R p 2 Two evaluation indexes, when R 2 and / or R p 2 reach the maximum, and R 2 and R p 2 are closest, the optimal number of components is obtained; wherein, R p 2 is determined by leave-one-out cross-validation ； ； determining a PLSR regression model of the final lithium geochemical background values according to the optimal number of component variables.

5. A system for identifying lithium geochemical anomalies in a lithologically complex area for implementing the method of identifying lithium geochemical anomalies according to any one of claims 1 to 4, characterized in that, The method comprises: a data preprocessing module, configured to obtain geochemical data of a region and preprocess the geochemical data of the region; an independent variable determination module, configured to determine PLSR independent variable indexes according to the preprocessed geochemical data; a regression model construction module, configured to construct a PLSR regression model according to the PLSR independent variable indexes; a background value determination module, configured to determine lithium geochemical background values of each sample point in the region according to the PLSR regression model; an anomaly identification module, configured to identify lithium geochemical anomalies in the region according to a predicted upper limit of the lithium geochemical background values.

Images