Ground mineral prediction method based on hyperspectral and geochemical data fusion

Abstract

The invention provides a ground mineral prediction method based on hyperspectral and geochemical data fusion, and the method comprises the steps: carrying out the preprocessing of hyperspectral remote sensing data and geochemical data, and carrying out the feature extraction of the hyperspectral remote sensing data and geochemical data; then, a proper data fusion method is selected to fuse the hyperspectral remote sensing data and the geochemical data, finally, a prediction model is established, and anomaly recognition and target area delineation are carried out. According to the method, hyperspectral and geochemical data can be conveniently fused, the advantages of the two methods are fully utilized, and the efficiency and accuracy of mineral resource exploration can be improved; the method provides support in the aspects of enhancing the recognition capability, expanding the detection range, improving the prediction precision, reducing the exploration cost, increasing the exploration speed, improving the prospecting success rate and supporting sustainable development.

Description

Technical Field

[0001] The present invention belongs to the field of mineral prediction, and in particular relates to a ground mineral prediction method based on the fusion of hyperspectral and geochemical data. Background Art

[0002] Hyperspectral remote sensing is an advanced remote sensing technology that identifies and analyzes surface materials by acquiring the reflectance or radiation characteristics of surface objects in continuous spectral bands. This technology can detect very subtle spectral differences to distinguish different rocks, minerals, vegetation, etc. In mineral prediction, hyperspectral remote sensing technology can be used to identify altered minerals and rock types associated with mineralization, because these materials may show unique characteristics in the spectrum. For example, by analyzing aerial hyperspectral remote sensing data from Liuyuan-Fangshankou in Beishan, Gansu and Xuemistan area in Xinjiang, researchers were able to identify the presence of solid minerals and established prediction methods such as mineralization environment analysis method, ore deposit location model identification method and mineralization structure tracking method based on aerial hyperspectral remote sensing.

[0003] Geochemical mineral prediction focuses on studying the distribution patterns of geochemical elements on the surface or underground. By analyzing the chemical elements in samples such as soil, stream sediments, and rocks, geochemical anomalies related to mineralization can be found. These anomalies may be element enrichment or dispersion caused by the presence of ore bodies. Geochemical survey data not only contains anomaly information related to mineralization, but also contains rich geological information, such as the distribution of geological units, the phase or lithofacies division of intrusive units, etc. This information provides a basis for the attribution of geological units and can successfully guide subsequent geological surveys. In addition, the application of geochemical data in the analysis of geological background and mineralization conditions is also very important, especially in complex terrain areas such as the arid desert area in northwest China. Geochemical survey data can effectively identify the distribution of geological units and provide important clues for prospecting prediction.

[0004] Hyperspectral technology and geochemical methods each have their own advantages in mineral prediction. They can complement each other and be used in combination to improve the success rate and efficiency of prospecting. Hyperspectral technology has the advantages of high efficiency and high precision in mineral identification and altered mineral analysis, while geochemical methods play an important role in providing geological background information and discovering geochemical anomalies. The combined use of the two can provide more comprehensive and in-depth geological information for mineral exploration.

[0005] The existing technologies mainly have the following disadvantages or shortcomings:

[0006] (1) Single data source

[0007] Traditional mineral exploration may rely on a single data source, such as using only geochemical data or remote sensing data, which may result in incomplete information;

[0008] (2) Limited prediction accuracy

[0009] A single data source may not provide sufficient information to accurately predict the distribution of a mineral resource;

[0010] (3) High cost

[0011] Traditional field exploration methods are costly and inefficient;

[0012] (4) Limited detection range

[0013] A single data source may not cover a large area, limiting the breadth of detection;

[0014] (5) Complex data processing

[0015] Traditional data processing methods may be complex and require a lot of manual intervention;

[0016] (6) Slow update speed

[0017] Traditional exploration methods update data slowly but cannot reflect geological changes in a timely manner.

[0018] To sum up, it is necessary to make further innovations on existing technologies. Summary of the invention

[0019] In response to the technical problems existing in the above-mentioned background technology, the present invention proposes a ground mineral prediction method based on the fusion of hyperspectral and geochemical data. The method has a reasonable conception and can conveniently integrate hyperspectral and geochemical data, making full use of the advantages of the two methods, thereby improving the efficiency and accuracy of mineral resource exploration, and providing support in enhancing recognition capabilities, expanding detection range, improving prediction accuracy, reducing exploration costs, accelerating exploration speed, increasing the success rate of prospecting and supporting sustainable development.

[0020] To solve the above technical problems, the present invention provides a ground mineral prediction method based on the fusion of hyperspectral and geochemical data, which first pre-processes the hyperspectral remote sensing data and geochemical data respectively, then extracts the features of the hyperspectral remote sensing data and geochemical data, then selects a suitable data fusion method to fuse the hyperspectral remote sensing data and geochemical data, and finally establishes a prediction model to carry out anomaly identification and target area delineation.

[0021] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein the specific process of preprocessing the hyperspectral remote sensing data and geochemical data respectively is: collecting hyperspectral remote sensing data of the working area and collecting geochemical samples; then completing radiation correction, atmospheric correction and geometric correction of the hyperspectral remote sensing data to ensure data quality, performing element content analysis on the geochemical samples, and carrying out data cleaning, standardization and normalization processing.

[0022] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein the specific process of extracting features from hyperspectral remote sensing data and geochemical data is: extracting mineral spectral features from hyperspectral remote sensing data to identify altered minerals related to mineralization; extracting element anomaly features from geochemical data to identify element combinations that may indicate mineralization.

[0023] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein the specific process of fusing the hyperspectral remote sensing data and the geochemical data is as follows: the hyperspectral remote sensing data is continuous data, which can be approximated as a Gaussian distribution to calculate the statistical characteristics of the target element; the geochemical anomaly is a group of element combinations, which belongs to classified data, and the elements can be numbered through the Markov random field, the element label random field is established, and the Gibbs distribution is established. The distribution form of the Gibbs is:

[0024]

[0025] In the above formula, Z is a normalization constant, T represents temperature, and represents the energy function, also called the potential function. Gaussian distribution and Gibbs distribution are unified into the Bayesian framework to construct the posterior probability distribution function:

[0026] p(m|d)=const×p(d|m)p(m);

[0027] In the above formula, p(d|m) is the conditional probability of d for m, and p(m) is the marginal probability of m. The prior information is the product of the Gaussian distribution of the hyperspectral data and the Gibbs distribution of the geochemical data, which can be expressed as:

[0028] p(m)=p1(m)p2(m)….

[0029] The method for predicting ground mineral resources based on the fusion of hyperspectral and geochemical data, wherein the process of approximating the hyperspectral remote sensing data to a Gaussian distribution is as follows: let d represent the data vector, m represent the model parameter vector; assume that the theoretical relationship between the data and the model parameters is represented by d=g(m); the observed value of the true value of d is represented by d obs Indicates that the prior information of m is expressed as m pri express;

[0030] It is assumed that the errors are independently distributed and conform to the zero-mean Gaussian distribution, with standard deviations of σ d and σ m , then the posterior probability distribution can be expressed as follows:

[0031]

[0032] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein the specific process of establishing the prediction model is: selecting a suitable machine learning algorithm to construct a prediction model, that is, a combination model of a convolutional neural network and a long short-term memory network, and then tuning the model parameters to find the optimal parameter combination to improve the prediction performance of the model.

[0033] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein: the machine learning algorithm includes support vector machine, random forest, neural network and LightGBM.

[0034] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein the method for tuning the model parameters adopts Bayesian optimization, and the specific process of using Bayesian optimization to tune the parameters of the combined model of convolutional neural network and long short-term memory network is as follows:

[0035] (4.01) Define optimization goals

[0036] Define an objective function, take the validation set loss or validation set accuracy of the combined model of the convolutional neural network and the long short-term memory network as the optimization target, and determine the parameters to be optimized and their value ranges;

[0037] (4.02) Select the Bayesian optimization library;

[0038] (4.03) Preparing the model

[0039] Define the combined model architecture of convolutional neural network and long short-term memory network, and prepare training data and validation data;

[0040] (4.04) Define the hyperparameter search space

[0041] Define convolutional neural network parameters, long short-term memory network parameters and training parameters;

[0042] (4.05) Configuring Bayesian Optimization

[0043] Set up a random exploration step to obtain initial data points, choose an acquisition function to guide subsequent exploration, and choose a Bayesian optimization algorithm;

[0044] (4.06) Run Bayesian Optimization

[0045] Use the Bayesian optimization algorithm to find the optimal parameters in the defined search space, set the number of iterations, and each iteration will select new parameters for trial based on the acquisition function;

[0046] (4.07) Model training and evaluation

[0047] Use the parameter training model recommended by Bayesian optimization and evaluate the performance of the parameter training model on the validation set;

[0048] (4.08) Result Analysis

[0049] Record and analyze the optimal parameters found by Bayesian optimization, and compare the performance of the combined model of convolutional neural network and long short-term memory network under the optimal parameters with the baseline model;

[0050] (4.09) Model verification

[0051] Use the test set to verify the combined model performance of the convolutional neural network and long short-term memory network under the optimal parameters;

[0052] (4.10) Model deployment

[0053] The trained combined model of convolutional neural network and long short-term memory network is saved for subsequent use, and the combined model of convolutional neural network and long short-term memory network is applied to actual mineral prediction tasks.

[0054] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data, wherein the specific process of carrying out anomaly identification and target area delineation is: by fusing hyperspectral remote sensing data and geochemical data and establishing a prediction model, hyperspectral anomalies, geochemical anomalies and target element distribution probability maps are obtained; the prediction model is used to identify geochemical anomalies and hyperspectral anomaly areas;

[0055] In addition, a constraint condition must be added when applying the prediction model to identify geochemical anomalies and hyperspectral anomaly areas: it is assumed that the distribution of solid minerals within a certain scale has a certain continuity, and a Markov random field neighborhood system is used to constrain it. When determining whether a certain point is an anomaly, it is necessary to consider the combination relationship between hyperspectral and geochemical elements established by machine learning in the neighborhood system for comprehensive judgment; by fusing hyperspectral remote sensing data and geochemical data, the information of hyperspectral anomalies and geochemical anomalies is integrated to delineate the prospecting target area.

[0056] By adopting the above technical solution, the present invention has the following beneficial effects:

[0057] The ground mineral prediction method based on the fusion of hyperspectral and geochemical data of the present invention is reasonably conceived, specifically carries out prediction for solid minerals, uses hyperspectral and geochemical data to carry out data fusion, and fully utilizes their respective advantages to carry out mineral prediction work quickly, efficiently and accurately.

[0058] The present invention can conveniently integrate hyperspectral and geochemical data, fully utilize the advantages of the two methods, improve the efficiency and accuracy of mineral resource exploration, and provide support in enhancing recognition capabilities, expanding detection range, improving prediction accuracy, reducing exploration costs, accelerating exploration speed, increasing the success rate of prospecting, and supporting sustainable development.

[0059] The present invention uses the fusion of hyperspectral data and geochemical data to carry out solid mineral prediction, innovatively integrating two different types of data, because it can provide more comprehensive and accurate geological information, can improve the accuracy of mineral exploration; it makes full use of the low-cost exploration technology of hyperspectral to expand the detection range, and on this basis, it can make full use of the high precision of geochemical exploration methods to improve the success rate of mineral exploration. The present invention can effectively improve the prediction efficiency, reduce the exploration cost, and at the same time establish a new data integration method to improve the efficiency of data utilization.

[0060] The present invention aims to improve the efficiency and accuracy of mineral resource exploration by combining two different data sources, namely, hyperspectral remote sensing data and geochemical data; compared with traditional technologies, the present invention mainly has the following characteristics or advantages:

[0061] (1) Improve prediction accuracy

[0062] By fusing hyperspectral remote sensing data with geochemical data, alteration minerals and element anomalies associated with mineralization can be more accurately identified, thereby improving the accuracy of mineral resource prediction;

[0063] (2) Expand the detection range

[0064] Hyperspectral remote sensing technology can cover larger surface areas, while geochemical data provides more detailed subsurface information; combining these two types of data can expand the detection range and cover a wider area;

[0065] (3) Reduce exploration costs

[0066] Traditional mineral exploration methods may require a lot of field exploration work, which is costly. By using the fusion technology of hyperspectral remote sensing and geochemical data, mineral prediction can be carried out without large-scale field exploration, thus reducing costs.

[0067] (4) Speed ​​up exploration

[0068] Traditional mineral exploration may take a long time to collect and analyze data; through automated data processing and fusion technology, the speed of mineral exploration can be accelerated;

[0069] (5) Improve the success rate of mineral exploration

[0070] The success rate of finding mineral resources can be improved through more accurate prediction models and anomaly identification technology;

[0071] (6) Supporting sustainable development

[0072] By improving exploration efficiency and accuracy, the development of mineral resources can be planned more rationally to support sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0074] Figure 1 Schematic diagram of the Markov random field neighborhood system involved in the ground mineral prediction method based on the fusion of hyperspectral and geochemical data of the present invention (Figure (a) is a first-order graph, and Figure (b) is a second-order graph);

[0075] Figure 2 It is a schematic diagram of Gaussian distribution involved in the ground mineral prediction method based on the fusion of hyperspectral and geochemical data of the present invention. DETAILED DESCRIPTION

[0076] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] The present invention is further explained below in conjunction with specific implementation modes.

[0078] like Figure 1As shown, the present embodiment provides a ground mineral prediction method based on the fusion of hyperspectral and geochemical data, which is to first pre-process the hyperspectral remote sensing data and the geochemical data respectively, then perform feature extraction of the hyperspectral remote sensing data and the geochemical data, and then select a suitable data fusion method to fuse the hyperspectral remote sensing data and the geochemical data, and finally establish a surface mineral prediction model to carry out mineral anomaly identification and mineral target area delineation.

[0079] The present invention provides a method for predicting ground mineral resources based on the fusion of hyperspectral and geochemical data, which specifically includes the following steps:

[0080] (1) Data collection and preprocessing

[0081] Collect hyperspectral remote sensing data and geochemical samples in the work area; then complete radiation correction, atmospheric correction and geometric correction (these correction methods are all conventional correction methods) on the hyperspectral remote sensing data to ensure data quality, analyze the element content of geochemical samples, and carry out data cleaning, standardization and normalization processing (all of which are conventional methods commonly used in big data mining).

[0082] (2) Feature extraction

[0083] Mineral spectral features are extracted from hyperspectral remote sensing data to identify altered minerals associated with mineralization; element anomaly features are extracted from geochemical data to identify element combinations that may indicate mineralization.

[0084] (3) Fusion of two types of data

[0085] Hyperspectral remote sensing data is continuous data, which can be approximated as a Gaussian distribution to calculate the statistical characteristics of the target elements;

[0086] The process of approximating the hyperspectral remote sensing data to a Gaussian distribution is as follows: let d represent the data vector and m represent the model parameter vector; assume that the theoretical relationship between the data and the model parameters is represented by d = g(m); the observed value of the true value of d is represented by d obs Indicates that the prior information of m is expressed as m pri express;

[0087] Assume that the errors are independently distributed and all conform to a zero-mean Gaussian distribution (e.g. Figure 2 ), and the standard deviations are σ d and σ m , then the posterior probability distribution can be expressed as follows:

[0088]

[0089] Geochemical anomalies are a group of element combinations and belong to classified data. The elements can be numbered through the Markov random field, the element number random field can be established, and the Gibbs distribution can be established. The distribution form of Gibbs is:

[0090]

[0091] In the above formula, Z is a normalization constant, T represents temperature, and represents the energy function, also called the potential function. Gaussian distribution and Gibbs distribution are unified into the Bayesian framework to construct the posterior probability distribution function:

[0092] p(m|d)=const×p(d|m)p(m);

[0093] In the above formula, p(d|m) is the conditional probability of d for m, and p(m) is the marginal probability of m. The prior information is the product of the Gaussian distribution of the hyperspectral data and the Gibbs distribution of the geochemical data, which can be expressed as:

[0094] p(m)=p1(m)p2(m)…;

[0095] (4) Establishing a prediction model

[0096] Select a suitable machine learning algorithm to build a prediction model; commonly used machine learning algorithms include support vector machine (SVM), random forest, neural network and LightGBM, etc.; using the parameters of the above fusion process, select the combined model of convolutional neural network (CNN) and long short-term memory network (LSTM) supplemented above as the prediction model; use methods such as grid search, random search or Bayesian optimization to tune the model parameters and find the optimal parameter combination to improve the prediction performance of the model. Among them, Bayesian optimization is a very effective method for tuning model parameters, especially when dealing with complex models with a large number of parameters; the following is a specific process of using Bayesian optimization to tune the parameters of the combined model of convolutional neural network (CNN) and long short-term memory network (LSTM):

[0097] (4.01) Define optimization goals

[0098] Objective function: define an objective function, usually the validation set loss or validation set accuracy of the combined model of the convolutional neural network and the long short-term memory network, as the optimization target; parameter range: determine the parameters to be optimized and their value range.

[0099] (4.02) Select the Bayesian Optimization Library

[0100] Library selection: Choose a Bayesian optimization library such as GPyOpt, Hyperopt, or Scikit-optimize.

[0101] (4.03) Preparing the model

[0102] Model definition: define the combined model architecture of CNN and LSTM; Data preparation: prepare training data and validation data.

[0103] (4.04) Define the hyperparameter search space

[0104] CNN parameters: such as the number of convolutional layers, the number of filters in each layer, the filter size, the type of pooling layer, etc.; LSTM parameters: such as the number of LSTM units, the number of layers, etc.; training parameters: such as learning rate, batch size, regularization parameters, etc.

[0105] (4.05) Configuring Bayesian Optimization

[0106] Initial random exploration: set a certain number of random exploration steps to obtain initial data points; Acquisition function: select an acquisition function, such as expected improvement (EI), probability density on Gaussian process, etc., to guide subsequent exploration; Optimization algorithm: select a Bayesian optimization algorithm, such as Gaussian process (GP).

[0107] (4.06) Run Bayesian Optimization

[0108] Optimization process: Use the Bayesian optimization algorithm to find the optimal parameters in the defined search space; Number of iterations: Set the number of iterations, and each iteration will select new parameters for trial based on the acquisition function.

[0109] (4.07) Model training and evaluation

[0110] Training model: Use the parameters recommended by Bayesian optimization to train the model; Evaluating model: Evaluate the performance of the parameter training model on the validation set.

[0111] (4.08) Result Analysis

[0112] Optimal parameters: record and analyze the optimal parameters found by Bayesian optimization; Performance comparison: compare the performance of the combined model of convolutional neural network and long short-term memory network under the optimal parameters with the baseline model.

[0113] (4.09) Model verification

[0114] Test set verification: Use the test set to verify the combined model performance of the convolutional neural network and long short-term memory network under the optimal parameters.

[0115] (4.10) Model deployment

[0116] Model saving: Save the trained combined model of convolutional neural network and long short-term memory network for subsequent use; Model application: Apply the combined model of convolutional neural network and long short-term memory network to actual mineral prediction tasks.

[0117] (5) Abnormality identification and target area delineation

[0118] Through the above steps (3)-(4), three results are obtained: hyperspectral anomaly, geochemical anomaly, and target element distribution probability map; the prediction model is used to identify geochemical anomaly and hyperspectral anomaly areas; a constraint is added to the process of using the prediction model to identify geochemical anomaly and hyperspectral anomaly areas: it is assumed that the distribution of solid minerals within a certain scale has a certain continuity, and a Markov random field neighborhood system is used to constrain it; the Markov random field is an extension of the one-dimensional Markov process. For the underground profile on the two-dimensional plane, it can be regarded as a Markov random field:

[0119] Let S = {(i, j)|1≤i≤M,1≤j≤N} (M and N are the number of rows and columns in the discrete two-dimensional space) be a set of spatial locations defined on the underground profile, if the set δ = {δ(s)|s∈S} satisfies the following conditions:

[0120]

[0121] It is called the set of neighborhood systems defined on . In this case, s is the center point, and is the neighborhood point. The order of the neighborhood system is defined by the distance between the neighborhood point and the center point. The distance here is usually the Euclidean distance. Figure 1 As shown in the figure, whether each central point contains minerals or not is related to the points in its neighborhood system. Through this constraint, a certain degree of continuity can be guaranteed. When determining whether a point is an abnormal point, it is necessary to consider the combination relationship between hyperspectral and geochemical elements established by machine learning in the neighborhood system to make a comprehensive judgment; through the fusion of hyperspectral remote sensing data and geochemical data, the information of hyperspectral anomalies and geochemical anomalies is integrated to delineate the prospecting target area.

[0122] The present invention has a reasonable concept, carries out prediction for solid minerals, uses hyperspectral and geochemical data to carry out data fusion, makes full use of their respective advantages, can carry out mineral prediction work quickly and efficiently, and is suitable for promotion and application.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A ground mineral prediction method based on the fusion of hyperspectral and geochemical data, characterized by: The method is to preprocess the hyperspectral remote sensing data and geochemical data respectively first, then extract the features of the hyperspectral remote sensing data and geochemical data, then select a suitable data fusion method to fuse the hyperspectral remote sensing data and geochemical data, and finally establish a prediction model to carry out anomaly identification and target area delineation.

2. The method for ground mineral prediction based on hyperspectral and geochemical data fusion as claimed in claim 1, characterized in that: The specific process of preprocessing the hyperspectral remote sensing data and geochemical data respectively is: collecting hyperspectral remote sensing data in the working area and collecting geochemical samples; then completing radiation correction, atmospheric correction and geometric correction for the hyperspectral remote sensing data to ensure data quality, performing element content analysis on the geochemical samples, and carrying out data cleaning, standardization and normalization processing.

3. The method for ground mineral prediction based on fusion of hyperspectral and geochemical data as claimed in claim 1, characterized in that: The specific process of extracting features from hyperspectral remote sensing data and geochemical data is as follows: extracting mineral spectral features from hyperspectral remote sensing data to identify altered minerals associated with mineralization; extracting element anomaly features from geochemical data to identify element combinations that may indicate mineralization.

4. The method for ground mineral prediction based on hyperspectral and geochemical data fusion as claimed in claim 1, characterized in that: The specific process of fusing the hyperspectral remote sensing data and geochemical data is as follows: the hyperspectral remote sensing data is continuous data, which can be approximated as a Gaussian distribution to calculate the statistical characteristics of the target element; the geochemical anomaly is a group of element combinations, which belongs to classified data. The elements can be numbered through the Markov random field, the element number random field is established, and the Gibbs distribution is established. The distribution form of the Gibbs is: In the above formula, Z is a normalization constant, T represents temperature, and represents the energy function, also called the potential function. Gaussian distribution and Gibbs distribution are unified into the Bayesian framework to construct the posterior probability distribution function: p(m|d)=const×p(d|m)p(m); In the above formula, p(d|m) is the conditional probability of d for m, and p(m) is the marginal probability of m. The prior information is the product of the Gaussian distribution of the hyperspectral data and the Gibbs distribution of the geochemical data, which can be expressed as: p(m)=p1(m) p2(m)….

5. The method for ground mineral prediction based on fusion of hyperspectral and geochemical data as claimed in claim 4, characterized in that: The process of approximating the hyperspectral remote sensing data to a Gaussian distribution is as follows: let d represent the data vector and m represent the model parameter vector; assume that the theoretical relationship between the data and the model parameters is represented by d=g(m); the observed value of the true value of d is represented by d obs Indicates that the prior information of m is expressed as m pri express; It is assumed that the errors are independently distributed and conform to the zero-mean Gaussian distribution, with standard deviations of σ d and σ m , then the posterior probability distribution can be expressed as follows:

6. The method for ground mineral prediction based on fusion of hyperspectral and geochemical data as claimed in claim 1, characterized in that: The specific process of establishing the prediction model is: selecting a suitable machine learning algorithm to construct a prediction model, that is, a combination model of a convolutional neural network and a long short-term memory network, and then tuning the model parameters to find the optimal parameter combination to improve the prediction performance of the model.

7. The method for ground mineral prediction based on the fusion of hyperspectral and geochemical data as claimed in claim 6, characterized in that: The machine learning algorithms include support vector machine, random forest, neural network and LightGBM.

8. The method for ground mineral prediction based on fusion of hyperspectral and geochemical data as claimed in claim 6, characterized in that: The method for tuning the model parameters adopts Bayesian optimization, and the specific process of using Bayesian optimization to tune the parameters of the combined model of the convolutional neural network and the long short-term memory network is as follows: (4.01) Define optimization goals Define an objective function, take the validation set loss or validation set accuracy of the combined model of the convolutional neural network and the long short-term memory network as the optimization target, and determine the parameters to be optimized and their value ranges; (4.02) Select the Bayesian optimization library; (4.03) Prepare the model Define the combined model architecture of convolutional neural network and long short-term memory network, and prepare training data and validation data; (4.04) Define the hyperparameter search space Define convolutional neural network parameters, long short-term memory network parameters and training parameters; (4.05) Configuring Bayesian Optimization Set up a random exploration step to obtain initial data points, choose an acquisition function to guide subsequent exploration, and choose a Bayesian optimization algorithm; (4.06) Run Bayesian Optimization Use the Bayesian optimization algorithm to find the optimal parameters in the defined search space, set the number of iterations, and each iteration will select new parameters for trial based on the acquisition function; (4.07) Model training and evaluation Use the parameter training model recommended by Bayesian optimization and evaluate the performance of the parameter training model on the validation set; (4.08) Result Analysis Record and analyze the optimal parameters found by Bayesian optimization, and compare the performance of the combined model of convolutional neural network and long short-term memory network under the optimal parameters with the baseline model; (4.09) Model verification Use the test set to verify the combined model performance of the convolutional neural network and long short-term memory network under the optimal parameters; (4.10) Model deployment The trained combined model of convolutional neural network and long short-term memory network is saved for subsequent use, and the combined model of convolutional neural network and long short-term memory network is applied to actual mineral prediction tasks.

9. The method for ground mineral prediction based on hyperspectral and geochemical data fusion as claimed in claim 1, characterized in that: The specific process of carrying out anomaly identification and target area delineation is as follows: by fusing hyperspectral remote sensing data and geochemical data and establishing a prediction model, a hyperspectral anomaly, geochemical anomaly and target element distribution probability map is obtained; the prediction model is used to identify geochemical anomalies and hyperspectral anomaly areas; In addition, a constraint condition must be added when applying the prediction model to identify geochemical anomalies and hyperspectral anomaly areas: it is assumed that the distribution of solid minerals within a certain scale has a certain continuity, and a Markov random field neighborhood system is used to constrain it. When determining whether a certain point is an anomaly, it is necessary to consider the combination relationship between hyperspectral and geochemical elements established by machine learning in the neighborhood system for comprehensive judgment; by fusing hyperspectral remote sensing data and geochemical data, the information of hyperspectral anomalies and geochemical anomalies is integrated to delineate the prospecting target area.

Images