A rock weathering velocity determination method and system based on Bayesian algorithm

By combining GIS technology and Bayesian algorithm, the problem of low efficiency of traditional rock weathering speed measurement methods is solved, and more efficient and accurate measurement is achieved, reducing costs and improving the degree of automation.

CN118351966BInactive Publication Date: 2024-11-01INST OF KARST GEOLOGY CAGS +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410568918.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-09
Publication Date
2024-11-01
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional rock weathering speed measurement methods take a long time and are inefficient, and rely on on-site observation and human factors, making it difficult to achieve efficient and accurate measurements.

Method used

Combining GIS technology and Bayesian algorithm, by obtaining the historical observation data and geographical location of rocks, performing data preprocessing and spatial distribution information analysis, establishing a data matrix and training a regression model, and using Bayesian algorithm to establish a prediction model to measure weathering speed .

Benefits of technology

It improves the accuracy and efficiency of the determination of rock weathering speed, reduces the dependence on on-site observations, reduces manpower and material costs, and improves the degree of automation of data processing and the reliability of predicted results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118351966B_ABST
    Figure CN118351966B_ABST
Patent Text Reader

Abstract

The present invention discloses a rock weathering velocity determination method and system based on Bayesian algorithm, including obtaining historical observation data and geographical location of rock, the historical observation data including weathering depth, rock type and environmental data, calculating weathering velocity according to weathering depth, preprocessing the historical observation data; obtaining first data according to the correlation degree of historical observation data to weathering velocity, using GIS system to obtain spatial distribution information of rock according to geographical information, obtaining second data according to spatial distribution information and spatial correlation of weathering velocity; reducing multicollinearity of first data by second data to obtain influence data, using Bayesian algorithm to establish prediction model according to influence data and weathering velocity; inputting data to be detected into prediction model, obtaining weathering velocity determination result. The present invention adopts GIS system and Bayesian algorithm to accurately determine rock weathering velocity, reduces on-site observation, and significantly reduces cost.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the fields of geology and environmental science, and in particular to a rock weathering velocity determination method and system based on a Bayesian algorithm. Background Art

[0002] The determination of rock weathering rate is of great significance to the fields of geological research, engineering construction, environmental protection, etc. Traditional rock weathering rate determination methods mostly rely on field observations and empirical estimates, which are often time-consuming, inefficient, and limited by human factors and environmental conditions. With the development of GIS technology and Bayesian algorithm, researchers have begun to explore more accurate and efficient rock weathering rate determination methods. GIS technology can provide spatial distribution information of rocks, while Bayesian algorithm can perform probabilistic inference under uncertain environments. The combination of the two provides a new technical approach for the accurate determination of rock weathering rate. It is based on this background that the present invention aims to propose a new rock weathering rate determination method by combining GIS technology and Bayesian algorithm, in order to improve the accuracy and efficiency of the determination, reduce the dependence on field observations, reduce the cost of manpower and material resources, and improve the degree of automation of data processing and the reliability of prediction results. Summary of the invention

[0003] The purpose of the present invention is to provide a rock weathering velocity determination method and system based on Bayesian algorithm.

[0004] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0005] The present invention comprises the following steps:

[0006] Obtaining historical observation data and geographical location of the rock, the historical observation data including weathering depth, rock type and environmental data, calculating weathering speed according to the weathering depth, and preprocessing the historical observation data;

[0007] Obtaining first data according to the correlation between the historical observation data and the weathering speed, obtaining spatial distribution information of the rock according to the geographic information using a GIS system, and obtaining second data according to the spatial correlation between the spatial distribution information and the weathering speed;

[0008] The multicollinearity of the first data is reduced by the second data to obtain the influence data.

[0009] A data matrix is ​​established using the first data and the second data, wherein the rows of the data matrix correspond to the first data, and the columns of the data matrix correspond to the second data, the data matrix is ​​divided into a training matrix and a test matrix, and a regression model is trained using the training matrix, and the loss function of the regression model is:

[0010]

[0011] Where v is the parameter vector of the regression model, t j is the first data, J is the number of rows of the data matrix, z j is with t j The corresponding second data, λ 1 , 2 and λ 3 is the regularization parameter, ρ is the smoothing coefficient, ρ∈[0,1], F represents the norm, and the loss function is optimized:

[0012]

[0013] Where sign is the sign function, and the update formula of the parameter vector is:

[0014]

[0015] where v t represents the parameter vector of the t-th iteration, Represents the learning rate, with a value range of [0.001, 0.1]. The optimal parameter vector is:

[0016]

[0017] in represents the value of the parameter vector v when the loss function is minimum, uses the parameter vector to train the regression model, uses the test matrix to verify the accuracy of the regression model, removes the first data whose absolute value of the parameter vector is less than a preset parameter threshold, the preset parameter threshold range is [0.1, 0.5], and uses the second data and the first data after screening as the influencing data;

[0018] Establishing a prediction model based on the impact data and the weathering rate using a Bayesian algorithm;

[0019] The data to be detected is input into the prediction model to obtain the measurement result of the weathering speed.

[0020] Furthermore, the method of calculating the weathering speed according to the weathering depth and preprocessing the historical observation data includes:

[0021] The change of the weathering depth per unit time is taken as the weathering speed, and the rock type is feature extracted and converted into numerical data greater than 0 and less than 1, and the weathering depth and the environmental data are normalized and preprocessed.

[0022] Furthermore, the method for obtaining the first data according to the correlation between the historical observation data and the weathering speed includes:

[0023] Using a multiple linear regression model:

[0024]

[0025] where Y it is the weathering velocity of the i-th rock at observation time t, β 0 is the intercept term, α is the fixed effect coefficient of the weathering depth, I is the weathering depth, β k is the fixed effect coefficient of the kth environmental data, E is the environmental data, γ is the fixed effect coefficient of the rock type, R i is the rock type of the i-th rock, δ t is the time fixed effect, ∈ it is the error term, m is the number of variables in the environmental data, β 0 , α, β k , γ is determined by maximum likelihood estimation during the training process of the multivariate linear regression model, the fixed effect coefficient is taken as the correlation degree, the product of the historical observation data whose absolute value of the correlation degree is greater than or equal to a preset threshold and the corresponding correlation degree is taken as the first data, and the value range of the preset threshold is [0.1, 0.5].

[0026] Furthermore, the method of obtaining the spatial distribution information of the rock according to the geographic information using a GIS system comprises:

[0027] The geographical location is imported into a GIS system to perform spatial distribution visualization to obtain the spatial distribution information, which includes altitude, slope, the type of landform and the distribution position in the landform type.

[0028] Furthermore, the method for obtaining the second data according to the spatial distribution information and the spatial correlation of the weathering velocity includes:

[0029] The landform type is encoded using one-hot encoding, the distribution position is vectorized using TF-IDF, the feature vector is established using the average value of the weathering speed of the rock and the spatial distribution position, and cluster analysis is performed to form c clusters:

[0030]

[0031] Among them, τ U is the objective function, n is the number of rocks, η ijis the membership of the ith rock to the jth cluster, indicating the spatial correlation of the rock to the cluster, M is the fuzzy factor, 1<M<3, Λ j is the fuzzy penalty coefficient of the jth cluster, X i is the eigenvector of rock i, u j is the center point of cluster j, C j is the rock set in cluster j, h is the bandwidth parameter, 0.4<h<1,

[0032] The updating formula of the membership degree is:

[0033]

[0034] Among them, δ is used to ensure that the denominator is not zero, 0<δ<0.01, and the cluster center update formula is:

[0035]

[0036] in is the center point of cluster j after iteration s+1 steps, and the weighted average of the weathering velocity of the cluster and the second data are calculated:

[0037]

[0038] where Y j is the average weathering velocity of the jth cluster, is the average weathering rate of the i-th rock, Y ib The second data is the weathering rate predicted based on the geographical information of the rock.

[0039] Furthermore, a method for establishing a prediction model based on the impact data and the weathering speed using a Bayesian algorithm includes:

[0040] The influencing data and the corresponding weathering speed are divided into a training set and a test set, wherein the weathering speed is used as a label of the influencing data, and the prediction model is trained using the training set:

[0041] The expected prediction results of the prediction model are:

[0042] Ψ(Y|X,D)=∫Ψ(Y|X,W)Ψ(W|D)dW

[0043] Where Ψ(Y|X,D) represents the probability distribution of the prediction result Y obtained by inputting the impact data X under the training set D, Y is the weathering speed in the training set D, W is the parameter set of the prediction model, and the goodness of fit of the prediction model is:

[0044] F(D,μ,θ)=∑lnq(w i |θ,γ)-μ∑ln[Π(w i )P(w i )]-∑lnΨ(D|w i ,γ)

[0045] Where D is the training set, θ is the parameter of the prediction model, q represents the variational distribution, and w i is the sampling value of the weight distribution of each neuron, μ is the weight of the prior distribution and the likelihood function term in the goodness of fit, Π(w i ) is w i The prior distribution of is a uniform prior, P(w i ) is w i The likelihood function, γ is the uncertainty parameter, and the probability density function is:

[0046]

[0047] where Y NN is the predicted value of the prediction model, α is the adjustment function, which represents the influence strength of the neighboring nodes on the target node. Substituting the probability density function into the goodness of fit, the goodness of fit of the prediction uncertainty is obtained:

[0048]

[0049] The test set is used to evaluate the generalization ability of the trained prediction model.

[0050] A rock weathering velocity determination system based on Bayesian algorithm, characterized by comprising:

[0051] A collection module, which obtains historical observation data and geographical location of rocks, wherein the historical observation data includes weathering depth, rock type and environmental data;

[0052] A calculation module, preprocessing the historical observation data, calculating the weathering speed according to the weathering depth, obtaining first data according to the correlation between the historical observation data and the weathering speed, obtaining spatial distribution information of the rock according to the geographic information using a GIS system, obtaining second data according to the spatial correlation between the spatial distribution information and the weathering speed, and establishing a prediction model according to the impact data and the weathering speed using a Bayesian algorithm;

[0053] An optimization module, which reduces the multicollinearity of the first data by using the second data to obtain influencing data;

[0054] The output module inputs the data to be detected into the prediction model to obtain the measurement result of the weathering speed.

[0055] The beneficial effects of the present invention are:

[0056] The present invention can effectively improve the accuracy and efficiency of rock weathering rate determination by combining GIS technology and Bayesian algorithm. The present invention not only reduces the dependence on field observation and reduces the manpower and material costs, but also improves the automation of data processing and the reliability of prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 The present invention is a flow chart of the steps of a rock weathering velocity determination method and system based on a Bayesian algorithm. DETAILED DESCRIPTION

[0058] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0059] The present invention provides a rock weathering velocity determination method and system based on Bayesian algorithm, comprising the following steps:

[0060] like Figure 1 As shown, in this embodiment, the following steps are included:

[0061] Obtaining historical observation data and geographical location of the rock, the historical observation data including weathering depth, rock type and environmental data, calculating weathering speed according to the weathering depth, and preprocessing the historical observation data;

[0062] Obtaining first data according to the correlation between the historical observation data and the weathering speed, obtaining spatial distribution information of the rock according to the geographic information using a GIS system, and obtaining second data according to the spatial correlation between the spatial distribution information and the weathering speed;

[0063] The multicollinearity of the first data is reduced by the second data to obtain the influence data.

[0064] A data matrix is ​​established using the first data and the second data, wherein the rows of the data matrix correspond to the first data, and the columns of the data matrix correspond to the second data, the data matrix is ​​divided into a training matrix and a test matrix, and a regression model is trained using the training matrix, and the loss function of the regression model is:

[0065]

[0066] Where v is the parameter vector of the regression model, t j is the first data, J is the number of rows of the data matrix, z j is with t j The corresponding second data, λ1 , 2 and λ 3 is the regularization parameter, ρ is the smoothing coefficient, ρ∈[0,1], the value is 0.5, F represents the norm, and the loss function is optimized:

[0067]

[0068] Where sign is the sign function, and the update formula of the parameter vector is:

[0069]

[0070] where v t represents the parameter vector of the t-th iteration, Represents the learning rate, the value range is [0.001, 0.1]. The optimal parameter vector:

[0071]

[0072] in represents the value of the parameter vector v when the loss function is minimum, uses the parameter vector to train the regression model, uses the test matrix to verify the accuracy of the regression model, removes the first data whose absolute value of the parameter vector is less than a preset parameter threshold, the preset parameter threshold range is [0.1, 0.5], the preset parameter threshold is 0.1, and uses the second data and the first data after screening as the influencing data;

[0073] Establishing a prediction model based on the impact data and the weathering rate using a Bayesian algorithm;

[0074] The data to be detected is input into the prediction model to obtain the measurement result of the weathering speed.

[0075] In this embodiment, the historical observation data obtained is:

[0076]

[0077] The specific observation time of the weathering depth is December 31st of each year, and the geographical locations obtained are: R1 (40.3N, 80.1W), R2 (40.4N, 80.2W), and R3 (40.2N, 80.0W).

[0078] In this embodiment, the method of calculating the weathering speed according to the weathering depth and preprocessing the historical observation data includes:

[0079] The change in the weathering depth per unit time is taken as the weathering speed:

[0080]

[0081] Where D is the weathering depth, Δt is the observation time interval,

[0082] The rock type is feature extracted and converted into numerical data greater than 0 and less than 1, the Mohs hardness of the rock type is divided by 10 to obtain the numerical data, and the weathering depth and the environmental data are normalized and preprocessed.

[0083] In this embodiment, the weathering rate is: R1: 2022, 2 mm / year, 2023, 2 mm / year; R2: 2022, 2 mm / year, 2023, 1 mm / year; R3: 2022, 2 mm / year, 2023, 2 mm / year; the standardized historical observation data:

[0084]

[0085]

[0086] In this embodiment, the method for obtaining the first data according to the correlation between the historical observation data and the weathering speed includes:

[0087] Using a multiple linear regression model:

[0088]

[0089] where Y it is the weathering velocity of the i-th rock at observation time t, β 0 is the intercept term, α is the fixed effect coefficient of the weathering depth, I is the weathering depth, β k is the fixed effect coefficient of the kth environmental data, E is the environmental data, γ is the fixed effect coefficient of the rock type, R i is the rock type of the i-th rock, δ t is the time fixed effect, ∈ it is the error term, m is the number of variables in the environmental data, β 0 , α, β k , γ is determined by maximum likelihood estimation during the training of the multivariate linear regression model, β 0 =1,δ t =0.1, the fixed effect coefficient is taken as the correlation degree, and the product of the historical observation data whose absolute value of the correlation degree is greater than or equal to a preset threshold and the corresponding correlation degree is taken as the first data, the value range of the preset threshold is [0.1, 0.5], and the preset threshold is 0.1.

[0090] In this embodiment, the correlation degree is: α=0.5, β k ={0.1, 0.2, 0.3} (corresponding to the annual average temperature, annual average humidity and annual average precipitation), γ = 0.2; the first data is:

[0091]

[0092] In this embodiment, the method of obtaining the spatial distribution information of the rock according to the geographic information using the GIS system includes:

[0093] The geographical location is imported into a GIS system to perform spatial distribution visualization to obtain the spatial distribution information, which includes altitude, slope, the type of landform and the distribution position in the landform type.

[0094] Rock Number Landform type Distribution location Altitude(m) Slope(°) R1 hills Gentle slope at the top of the hill 300.00 10.00 R2 Plains Low-lying areas of the plain 150.00 3.00 R3 River Valley Floodplain area on the north bank of the river 95.00 5.00

[0095] In this embodiment, the method for obtaining the second data according to the spatial distribution information and the spatial correlation of the weathering velocity includes:

[0096] The landform type is encoded using one-hot encoding, the distribution position is vectorized using TF-IDF, and the feature vector is established using the average weathering speed and spatial distribution position of the rock. The average weathering speed of the rock is:

[0097]

[0098] Where T is the number of weathering speeds of the rock, cluster analysis is performed on the feature vector to form c clusters, where c=3:

[0099]

[0100] Among them, τ U is the objective function, n is the number of rocks, η ij is the membership of the ith rock to the jth cluster, indicating the spatial correlation of the rock to the cluster, M is the fuzzy factor, 1<M<3, M=1.5, Λ j is the fuzzy penalty coefficient of the jth cluster, X i is the eigenvector of rock i, u j is the center point of cluster j, C j is the rock set in cluster j, h is the bandwidth parameter, 0.4<h<1, and the value is 0.5.

[0101] The updating formula of the membership degree is:

[0102]

[0103] Among them, δ is used to ensure that the denominator is not zero, 0<δ<0.01, and the cluster center update formula is:

[0104]

[0105] in is the center point of cluster j after iteration s+1 steps, and the weighted average of the weathering velocity of the cluster and the second data are calculated:

[0106]

[0107] where Y j is the average weathering velocity of the jth cluster, is the average weathering rate of the i-th rock, Y ib The second data is the weathering rate predicted based on the geographical information of the rock.

[0108] In this embodiment, the second data is: R1: 2 mm / year; R2: 1.5 mm / year; R3: 2 mm / year.

[0109] In this embodiment, the impact data is:

[0110]

[0111]

[0112] In this embodiment, the method of establishing a prediction model according to the influencing data and the weathering speed using the Bayesian algorithm includes:

[0113] The influencing data and the corresponding weathering speed are divided into a training set and a test set, wherein the weathering speed is used as a label of the influencing data, and the prediction model is trained using the training set:

[0114] The expected prediction results of the prediction model are:

[0115] Ψ(Y|X,D)=∫Ψ(Y|X,W)Ψ(W|D)dW

[0116] Where Ψ(Y|X,D) represents the probability distribution of the prediction result Y obtained by inputting the impact data X under the training set D, Y is the weathering speed in the training set D, W is the parameter set of the prediction model, and the goodness of fit of the prediction model is:

[0117] F(D,μ,θ)=∑lnq(w i |θ,γ)-μ∑ln[Π(w i )P(wi )]-∑lnΨ(D|w i ,γ)

[0118] Where D is the training set, θ is the parameter of the prediction model, q represents the variational distribution, and w i is the sampling value of the weight distribution of each neuron, μ is the weight of the prior distribution and the likelihood function term in the goodness of fit, Π(w i ) is w i The prior distribution of is a uniform prior, P(w i ) is w i The likelihood function, γ is the uncertainty parameter, and the probability density function is:

[0119]

[0120] where Y NN is the predicted value of the prediction model, α is the adjustment function, which represents the influence strength of the neighboring nodes on the target node. Substituting the probability density function into the goodness of fit, the goodness of fit of the prediction uncertainty is obtained:

[0121]

[0122] The test set is used to evaluate the generalization ability of the trained prediction model, and the prediction accuracy is determined based on the mean square error (MSE) between the predicted value and the actual value. When MSE is less than 0.1, the prediction model is qualified.

[0123] In this embodiment, MSE=0.05<0.1, and the prediction model is qualified.

[0124] A rock weathering velocity determination system based on Bayesian algorithm, characterized by comprising:

[0125] A collection module, which obtains historical observation data and geographical location of rocks, wherein the historical observation data includes weathering depth, rock type and environmental data;

[0126] A calculation module, preprocessing the historical observation data, calculating the weathering speed according to the weathering depth, obtaining first data according to the correlation between the historical observation data and the weathering speed, obtaining spatial distribution information of the rock according to the geographic information using a GIS system, obtaining second data according to the spatial correlation between the spatial distribution information and the weathering speed, and establishing a prediction model according to the impact data and the weathering speed using a Bayesian algorithm;

[0127] An optimization module, which reduces the multicollinearity of the first data by using the second data to obtain influencing data;

[0128] The output module inputs the data to be detected into the prediction model to obtain the measurement result of the weathering speed.

[0129] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for determining rock weathering velocity based on Bayesian algorithm, characterized in that: The following steps are involved: Obtaining historical observation data and geographical location of the rock, the historical observation data including weathering depth, rock type and environmental data, calculating weathering speed according to the weathering depth, and preprocessing the historical observation data; Obtaining first data according to the correlation between the historical observation data and the weathering speed, obtaining spatial distribution information of the rock according to geographic information using a GIS system, and obtaining second data according to the spatial correlation between the spatial distribution information and the weathering speed; The multicollinearity of the first data is reduced by the second data to obtain the influence data. A data matrix is ​​established using the first data and the second data, wherein the rows of the data matrix correspond to the first data, and the columns of the data matrix correspond to the second data, the data matrix is ​​divided into a training matrix and a test matrix, and a regression model is trained using the training matrix, and the loss function of the regression model is: Where v is the parameter vector of the regression model, t j is the first data, J is the number of rows of the data matrix, z j is with t j Corresponding to the second data, λ1, λ2 and λ3 are regularization parameters, ρ is a smoothing coefficient, ρ∈[0,1], F represents the norm, and the loss function is optimized: Where sign is the sign function, and the update formula of the parameter vector is: where v t represents the parameter vector of the t-th iteration, Represents the learning rate, with a value range of [0.001, 0.1]. The optimal parameter vector is: in represents the value of the parameter vector v when the loss function is minimum, uses the parameter vector to train the regression model, uses the test matrix to verify the accuracy of the regression model, removes the first data whose absolute value of the parameter vector is less than a preset parameter threshold, the preset parameter threshold range is [0.1, 0.5], and uses the second data and the first data after screening as the influencing data; Establishing a prediction model based on the impact data and the weathering rate using a Bayesian algorithm; The data to be detected is input into the prediction model to obtain the measurement result of the weathering speed.

2. The rock weathering velocity determination method based on Bayesian algorithm according to claim 1, characterized in that: The method of calculating the weathering speed according to the weathering depth and preprocessing the historical observation data comprises: The change of the weathering depth per unit time is taken as the weathering speed, and the rock type is feature extracted and converted into numerical data greater than 0 and less than 1, and the weathering depth and the environmental data are normalized and preprocessed.

3. The rock weathering velocity determination method based on Bayesian algorithm according to claim 1 is characterized in that: The method for obtaining first data according to the correlation degree of the historical observation data to the weathering speed includes: Using a multiple linear regression model: where Y it is the weathering velocity of the i-th rock at observation time t, β0 is the intercept term, α is the fixed effect coefficient of the weathering depth, I is the weathering depth, β k is the fixed effect coefficient of the kth environmental data, E is the environmental data, γ' is the fixed effect coefficient of the rock type, R i is the rock type of the i-th rock, δ t is the time fixed effect, is the error term, m is the number of variables in the environmental data, β0, α, β k , γ' is determined by maximum likelihood estimation during the training process of the multivariate linear regression model, the fixed effect coefficient is used as the correlation degree, the product of the historical observation data whose absolute value of the correlation degree is greater than or equal to a preset threshold and the corresponding correlation degree is used as the first data, and the value range of the preset threshold is [0.1, 0.5].

4. The rock weathering velocity determination method based on Bayesian algorithm according to claim 1 is characterized in that: The method of obtaining the spatial distribution information of the rock according to geographic information using a GIS system comprises: The geographical location is imported into a GIS system to perform spatial distribution visualization to obtain the spatial distribution information, which includes altitude, slope, the type of landform and the distribution position in the landform type.

5. The rock weathering velocity determination method based on Bayesian algorithm according to claim 1 is characterized in that: The method of establishing a prediction model according to the impact data and the weathering speed using a Bayesian algorithm comprises: The influencing data and the corresponding weathering speed are divided into a training set and a test set, wherein the weathering speed is used as a label of the influencing data, and the prediction model is trained using the training set: The expected prediction results of the prediction model are: Ψ(Y|X,D)=∫Ψ(Y|X,W)Ψ(W|D)dW Where Ψ(Y|X,D) represents the probability distribution of the prediction result Y obtained by inputting the impact data X under the training set D, Y is the weathering speed in the training set D, W is the parameter set of the prediction model, and the goodness of fit of the prediction model is: F(D,μ,θ)=∑lnq(w i ∣θ,γ)-μ∑ln[Π(w i )P(w i )]-∑lnΨ(D∣w i ,c) Where D is the training set, θ is the parameter of the prediction model, q represents the variational distribution, and w i is the sampling value of the weight distribution of each neuron, μ is the weight of the prior distribution and the likelihood function term in the goodness of fit, Π(w i ) is w i The prior distribution of is a uniform prior, P(w i ) is w i The likelihood function, γ is the uncertainty parameter, and the probability density function is: where Y NN is the predicted value of the prediction model, α is the adjustment function, which represents the influence strength of the neighboring nodes on the target node. Substituting the probability density function into the goodness of fit, the goodness of fit of the prediction uncertainty is obtained: The test set is used to evaluate the generalization ability of the trained prediction model.

6. A rock weathering rate measurement system based on Bayesian algorithm, used to execute the rock weathering rate measurement method based on Bayesian algorithm according to any one of claims 1 to 5, characterized in that: include: A collection module, which obtains historical observation data and geographical location of rocks, wherein the historical observation data includes weathering depth, rock type and environmental data; A calculation module, preprocessing the historical observation data, calculating the weathering speed according to the weathering depth, obtaining first data according to the correlation between the historical observation data and the weathering speed, obtaining spatial distribution information of the rock according to geographic information using a GIS system, obtaining second data according to the spatial correlation between the spatial distribution information and the weathering speed, and establishing a prediction model according to the impact data and the weathering speed using a Bayesian algorithm; An optimization module, which reduces the multicollinearity of the first data by using the second data to obtain influencing data; The output module inputs the data to be detected into the prediction model to obtain the measurement result of the weathering speed.