CN111680843B - Prediction method and system of Chinese herbal medicine suitable area based on deep SVDD model - Google Patents

Abstract

The invention discloses a Chinese medicinal material survival area prediction method and system based on a deep SVDD model, which comprises the following steps: collecting ecological factor data of the traditional Chinese medicinal materials, and generating pseudo-nonexistent sample data of the traditional Chinese medicinal materials by adopting a MaxEnt model; preprocessing the collected ecological factor data of the traditional Chinese medicinal materials to obtain ecological factor preprocessing data; constructing a prediction model of the Chinese medicinal material survival area according to the ecological factor preprocessing data, the pseudo-nonexistent sample data and the SVDD model; and putting the test points of the traditional Chinese medicinal materials to be predicted into the prediction model of the survival area of the traditional Chinese medicinal materials for judgment to obtain the survival area of the traditional Chinese medicinal materials to be predicted. The SGD and the SGD variant optimize parameters of a deep SVDD model, and the calculation complexity of the model is linear expansion on the training quantity, so that a large data set can be well expanded; the survival area is obtained by judging the distance between different test points and the optimal hypersphere, and the accuracy of the prediction result of the survival area of the traditional Chinese medicinal materials is improved.

Description

Prediction method and system of suitable growth area of Chinese medicinal materials based on deep SVDD model

technical field

The invention relates to the field of development and utilization of traditional Chinese medicine resources, in particular to a method and a system for predicting suitable growth areas of traditional Chinese medicinal materials based on a deep SVDD model.

Background technique

The development and utilization of traditional Chinese medicine resources and the protection of their sustainable development are very important for the research of traditional Chinese medicine resources in my country, because my country's traditional Chinese medicine resources face many problems. The effective components of medicinal materials are significantly different from the standard of "Chinese Pharmacopoeia", which seriously restricts the sustainable development of traditional Chinese medicinal materials. In order to more scientifically expand the areas where Chinese medicinal materials are introduced, it is necessary to strengthen the research on the ecological suitability of medicinal materials, find the ecological factors such as light, temperature, moisture, terrain, soil, etc. Rational use of environmental resources, protection of traditional Chinese medicine resources, to achieve the goal of sustainable development.

At present, most of the researches on the distribution prediction of the suitable areas for medicinal materials use the maximum entropy model and the existing distribution data and ecological environment to predict the distribution pattern and changes of the suitable areas. In the prior art, the method of combining the MaxEnt niche model and GIS technology was adopted. According to the distribution data of 214 Platycodon grandiflorum sample points, the contribution rate of ecological factors was analyzed based on the knife-cut method, and the main ecological factors and suitable habitat characteristics affecting the growth of Platycodon grandiflorum were explored. Therefore, the regionalization research of the suitable growth area of Platycodon grandiflorum in the whole country was carried out, and the prediction accuracy evaluation index AUC (Area Under Curve) value reached 0.922.

However, the Maxent model is a complex machine learning algorithm, which is sensitive to sampling bias and is prone to overfitting. The transfer ability of the Maxent model is only good in the case of a low threshold; and the Maxent model will be based on default parameters. have an impact on the accuracy of the prediction results.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is that the Maxent model is used in the existing traditional Chinese medicinal material suitable area prediction method, which leads to sampling deviation, which is prone to overfitting, and the Maxent model will affect the accuracy of the prediction result based on the default parameters. The purpose is to provide a method and system for predicting the suitable area of Chinese medicinal materials based on the deep SVDD model to solve the above problems.

The present invention is achieved through the following technical solutions:

A method for predicting suitable areas for Chinese medicinal materials based on a deep SVDD model, comprising:

S1: Collect ecological factor data of Chinese herbal medicines, and use MaxEnt model to generate pseudo-absence point sample data of Chinese herbal medicines;

The pseudo-absence point sample data is the unsuitable area of Chinese medicinal materials obtained by the MaxEnt model;

S2: preprocessing the collected ecological factor data of Chinese medicinal materials to obtain ecological factor preprocessing data;

S3: constructing a prediction model for the suitable area for Chinese medicinal materials according to the ecological factor preprocessing data, the pseudo-absence point sample data and the SVDD model;

S4: Put the test point of the Chinese herbal medicine to be predicted into the prediction model of the Chinese herbal medicine suitable area for judgment, and obtain the suitable area of the Chinese herbal medicine to be predicted.

The present invention proposes a prediction model for the suitable area of Chinese medicinal materials based on the deep support vector data description, that is, based on the deep SVDD model. Due to the different data formats of ecological environment factors, it is necessary to preprocess the data, that is, to realize the unified transformation of the data based on the t-SNE algorithm, and use the deep support vector data description model to map the transformed data to high-dimensional features based on nonlinearity. space, and find the optimal hypersphere in the feature space, using SGD and its variants to optimize the parameters of the deep SVDD model, since its computational complexity scales linearly in the number of training batches, it performs well on large datasets extension.

In the existing long-term research on Chinese medicinal materials resources, the country provides the storage and management of big data for the ecological factor data of Chinese medicinal materials, and is authoritative and authentic. The sample platform collects and obtains the sample distribution data of Chinese herbal medicines. Due to the wide distribution of sample points and the existence of duplicate data, it is necessary to screen the valid data by data cleaning and other methods to obtain the longitude and latitude of the sample points; by consulting the environmental factor literature of relevant Chinese herbal medicines , collect environmental factor data of traditional Chinese medicinal materials; the environmental factor data includes climate factor, terrain factor and soil factor; the longitude and latitude mapping is performed through the national basic geographic information system network to collect the ecological environment distribution in various regions of China.

In order to increase the reliability of the model, it is necessary to train the model together with sample data that does exist and does not exist, that is, a certain medicinal material must not grow in a certain place. Since there is no real nonexistent data, the present invention uses the optimized MaxEnt model after parameter adjustment to construct a false nonexistence data.

Since the input data is multi-source and heterogeneous, such as soil texture is text type, temperature, precipitation, etc. are numerical types, the present invention first uses the word vector model Word2vec to convert the words in the text into the representation of word vectors, and obtains the text data. Feature representation; due to the low processing efficiency of the word vector high-dimensional space, the t-SNE algorithm is used to map the high-dimensional word vector space into a two-dimensional space, so that two words with similar word meanings remain similar after mapping, and those with farther word meanings Words maintain a large mapping distance.

Further, the main ecological factor data for judging the suitable habitat of Chinese medicinal materials include sample distribution data, environmental factor data and map data.

Further, the step of generating the pseudo-absent point sample data in the S1 includes:

Use MaxEnt model to generate the numerical results of the suitable area of Chinese medicinal materials;

The output result of the MaxEnt model is 0 to 1, which means that each grid can be considered as the fitness index of the pixel in the map. The higher the value, the more suitable the Chinese medicinal material is. The grid is regarded as a suitable area. After the suitable area is selected, its latitude and longitude are removed from the map, leaving only the non-suitable area.

Eliminate the values that are greater than or equal to the threshold value in the result of the suitable area of the Chinese medicinal material to obtain the non-suitable area;

And the simulation effect of the model is the best when the number of existence points and pseudo-absence points is the same, so the present invention selects the same number of pseudo-absence points as the ecological factor data of Chinese medicinal materials from the non-suitable area to obtain the pseudo-absence of traditional Chinese medicinal materials. No point sample data exists.

Further, the preprocessing process of S2 includes:

The word vector model Word2vec is used to convert the ecological factor data into a high-dimensional space word vector;

Use the t-SNE algorithm to map high-dimensional spatial word vectors to two-dimensional spatial word vectors.

Further, the t-SNE algorithm:

In order to make similar objects have a higher probability to be selected, and dissimilar objects have a lower probability to be selected, the similarity between objects is expressed by converting Euclidean distance into conditional probability, that is, constructing a high-dimensional object between The probability distribution between different data, the similarity between different data is expressed:

Among them, p _j|i represents the similarity between different data in high-dimensional space, x _i and x _j are any two different data in the N-dimensional data x ₁ , x ₂ ,...,x _N , and the parameter σ _i represents the variance of the Gaussian distribution centered on x _i , and |||| represents the two-norm operation; the present invention only cares about the similarity between two different points, so set p _i|i =0;

Because it is necessary to map the vectors of the high-dimensional space to the low-dimensional space, in order to make the probability distribution of the same objects in the low-dimensional space as the high-dimensional space as similar as possible to the probability distribution of the high-dimensional space, it is necessary to map the high-dimensional space in the low-dimensional space to the probability distribution of the high-dimensional space. The dimensional object is used to construct the probability distribution, and the similarity between different data is expressed:

同理q_i|i＝0。Among them, q _j|i represents the similarity between different data in the low-dimensional space, and y _i and y _j represent the two-dimensional data y ₁ , y ₂ in the low-dimensional space; it is assumed that the Gaussian distribution has a variance of

Similarly q _i|i =0.

Construct the joint probability distributions P and Q of the high-dimensional space and the low-dimensional space, respectively, so that for any i and j, there are q _i|j =p _j|i , q _i|j =q _j|i .

Among them, p _i,j represents the joint probability between any two data in the high-dimensional space, and q _i,j represents the joint probability between any two data in the low-dimensional space.

Using KL divergence to measure the similarity of the joint probability distribution of high-dimensional space and low-dimensional space, we get:

Among them, C represents the similarity of the joint probability distribution of high-dimensional space and low-dimensional space, P represents the joint probability of high-dimensional space, and Q represents the joint probability of low-dimensional space.

Further, the construction process of the prediction model of the suitable area for Chinese medicinal materials in the S3:

将所述生态因子预处理数据映射到高维特征空间；The SVDD model uses a fully connected network

mapping the ecological factor preprocessing data to a high-dimensional feature space;

Find the optimal hypersphere in the high-level feature space, and the pseudo-absence point sample data is located outside the hypersphere of the optimal hypersphere, and the sample distribution data in the ecological factor data is located in the Inside the optimal hypersphere.

对于任意输入

第

层的输出为：Further, suppose

for any input

the first

The output of the layer is:

是第

层的激活函数，

是第

层的权重。where · represents a linear operation (e.g., matrix multiplication),

is the first

layer activation function,

is the first

layer weights.

The objective function is as follows:

到中心

的距离的平方小于半径二次方与松弛变量之和，其中n表示样本数大小，第二项是一个权重衰减，其采用L2正则化，其中λ为权重衰减系数，且λ>0，ξ_i是一个松弛变量，且满足

||·||_F为F-范数。因此，可以看作是发现一个以c为中心的最小体积超球面，通过最小化所有数据表示到中心的平均偏差来收缩球体半径。The first item in formula (10) is the sum of the squares of the radii and then the average value, and it satisfies each network representation

to the center

The square of the distance is less than the sum of the square of the radius and the slack variable, where n represents the number of samples, and the second term is a weight decay, which uses L2 regularization, where λ is the weight decay coefficient, and λ>0, ξ _i is a slack variable that satisfies

||·|| _F is the F-norm. Thus, it can be seen as finding a minimal volume hypersphere centered at c, shrinking the sphere radius by minimizing the mean deviation of all data representations from the center.

By minimizing formula (10), the Lagrangian multipliers α _i and β _i are introduced, and the Lagrangian function is constructed as follows:

stα _i ≥0,β _i ≥0

Taking the derivative of R, c, ξ, we can get:

Combining formula (11) and formula (12), we can get:

Then the formula for the radius and center of the optimal hypersphere is:

n表示样本数大小，

表示每个连接网络，α_i和α_j表示拉格朗日乘子，

表示内积，

表示支持向量，且

where c represents the center point, and

n is the sample size,

denote each connected network, α _i and α _j denote Lagrange multipliers,

represents the inner product,

represents the support vector, and

Further, in the described S4, select the test point of the Chinese medicinal material to be predicted in any of the ecological factor data;

Calculate the distance between the test point of the Chinese medicinal material to be predicted and the center point of the optimal hypersphere:

表示SVDD模型的超参数；Among them, x' represents the test point, s(x') represents the distance between the test point and the optimal hypersphere center point,

Represents the hyperparameters of the SVDD model;

Judging whether s(x') is greater than the radius of the hypersphere with the smallest volume, when s(x') is greater than the radius of the optimal hypersphere, the test point is an unsuitable area;

When s(x') is less than or equal to the radius of the optimal hypersphere, the test point is a suitable area;

The above operation is performed on all the test points of the Chinese herbal medicine to be predicted to obtain all the suitable areas of the Chinese herbal medicine to be predicted.

A prediction system based on the deep SVDD model for the suitable area of Chinese medicinal materials, including:

The collection module is used to collect ecological factor data of Chinese medicinal materials and generate pseudo-absence point sample data of Chinese medicinal materials;

The pseudo-absence point sample data is the unsuitable area of Chinese medicinal materials obtained by the MaxEnt model;

The preprocessing module is used for preprocessing the collected ecological factor data of Chinese medicinal materials to obtain ecological factor preprocessing data;

The prediction model generation module is used for constructing the prediction model of the suitable area for Chinese medicinal materials according to the ecological factor preprocessing data, the pseudo-absence point sample data and the SVDD model;

The prediction module is used to predict and obtain the suitable area of the Chinese herbal medicine to be predicted.

Further, the prediction process of the prediction module: select the test point of the Chinese medicinal material to be predicted in any of the ecological factor data;

Calculate the distance between the test point of the Chinese medicinal material to be predicted and the optimal hypersphere center point in the prediction model of the Chinese medicinal material suitable area:

表示SVDD模型的超参数；Among them, x' represents the test point, s(x') represents the distance between the test point and the center point of the hypersphere where the prediction model of the suitable area for Chinese medicinal materials is optimal,

Represents the hyperparameters of the SVDD model;

Judging whether s(x') is greater than the radius of the hypersphere with the smallest volume, when s(x') is greater than the radius of the optimal hypersphere, the test point is an unsuitable area;

When s(x') is less than or equal to the radius of the optimal hypersphere, the test point is a suitable area; perform the above operation on all the test points of the Chinese herbal medicine to be predicted to obtain all suitable conditions of the Chinese herbal medicine to be predicted. living area.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

1. The present invention is based on the deep SVDD model-based method and system for predicting the suitable areas for Chinese medicinal materials, using the SVDD model to train the suitable areas for Chinese medicinal materials, and using SGD and its variants to optimize the parameters of the deep SVDD model, because its computational complexity is The number of training batches scales linearly, so scales well for large datasets;

2. The present invention is based on the deep SVDD model for the prediction method and system of the suitable area for Chinese medicinal materials. When the SVDD model is used, it will not generate sampling deviation and over-fitting because it uses vector description for all data. The distance judgment of the hypersphere can obtain the suitable area, and improve the accuracy of the prediction result of the suitable area of Chinese medicinal materials.

Description of drawings

The accompanying drawings described herein are used to provide further understanding of the embodiments of the present invention, and constitute a part of the present application, and do not constitute limitations to the embodiments of the present invention. In the attached image:

Fig. 1 is the overall method flow chart of the present invention;

2 is a schematic diagram of the system structure of the present invention;

FIG. 3 is a schematic diagram of the operation of the SVDD model of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and the accompanying drawings. as a limitation of the present invention.

Example 1

As shown in Figure 1, a method for predicting the suitable area of Chinese medicinal materials based on the deep SVDD model includes:

S1: Collect ecological factor data of Chinese herbal medicines, and use MaxEnt model to generate pseudo-absence point sample data of Chinese herbal medicines;

The pseudo-absence point sample data is the unsuitable area of Chinese medicinal materials obtained by the MaxEnt model;

S2: preprocessing the collected ecological factor data of Chinese medicinal materials to obtain ecological factor preprocessing data;

S3: constructing a prediction model for the suitable area for Chinese medicinal materials according to the ecological factor preprocessing data, the pseudo-absence point sample data and the SVDD model;

S4: Put the test point of the Chinese herbal medicine to be predicted into the prediction model of the Chinese herbal medicine suitable area for judgment, and obtain the suitable area of the Chinese herbal medicine to be predicted.

The ecological factor data includes sample distribution data, environmental factor data and map data.

The step of generating the pseudo-absent point sample data in the S1 includes:

Use MaxEnt model to generate the numerical results of the suitable area of Chinese medicinal materials;

Eliminate the values that are greater than or equal to the threshold value in the result of the suitable area of the Chinese medicinal material to obtain the non-suitable area;

The same number of pseudo-absence points as the ecological factor data of Chinese medicinal materials are selected from the non-suitable area to obtain sample data of pseudo-absence points of traditional Chinese medicinal materials.

The preprocessing process of S2 includes:

The word vector model Word2vec is used to convert the ecological factor data into a high-dimensional space word vector;

Use the t-SNE algorithm to map high-dimensional spatial word vectors to two-dimensional spatial word vectors.

The t-SNE algorithm:

Construct a probability distribution between high-dimensional objects, and the similarity between different data is expressed as:

Among them, p _j|i represents the similarity between different data in high-dimensional space, x _i and x _j are any two different data in the N-dimensional data x ₁ , x ₂ ,...,x _N , and the parameter σ _i represents the variance of the Gaussian distribution centered on x _i , and |||| represents the two-norm operation;

The probability distribution is constructed for the high-dimensional objects in the low-dimensional space, and the similarity between different data is expressed as:

Among them, q _j|i represents the similarity between different data in the low-dimensional space, and y _i and y _j represent the two-dimensional data y ₁ , y ₂ in the low-dimensional space;

Construct the joint probability distributions P and Q of the high-dimensional space and the low-dimensional space, respectively, so that for any i and j, there are q _i|j =p _j|i , q _i|j =q _j|i .

Among them, p _i,j represents the joint probability between any two data in the high-dimensional space, and q _i,j represents the joint probability between any two data in the low-dimensional space.

Using KL divergence to measure the similarity of the joint probability distribution of high-dimensional space and low-dimensional space, we get:

Among them, C represents the similarity of the joint probability distribution of high-dimensional space and low-dimensional space, P represents the joint probability of high-dimensional space, and Q represents the joint probability of low-dimensional space.

The construction process of the prediction model of the suitable area for Chinese medicinal materials in the S3:

将所述生态因子预处理数据映射到高维特征空间；The SVDD model uses a fully connected network

mapping the ecological factor preprocessing data to a high-dimensional feature space;

Find the optimal hypersphere in the high-level feature space, and the pseudo-absence point sample data is located outside the hypersphere of the optimal hypersphere, and the sample distribution data in the ecological factor data is located in the Inside the optimal hypersphere.

The formula for the radius and center of the optimal hypersphere:

n表示样本数大小，

表示每个连接网络，α_i和α_j表示拉格朗日乘子，

表示内积，

表示支持向量，且

where c represents the center point, and

n is the sample size,

denote each connected network, α _i and α _j denote Lagrange multipliers,

represents the inner product,

represents the support vector, and

In described S4, select the test point of the Chinese medicinal material to be predicted in any of the ecological factor data;

Calculate the distance between the test point of the Chinese medicinal material to be predicted and the center point of the optimal hypersphere:

表示SVDD模型的超参数；Among them, x' represents the test point, s(x') represents the distance between the test point and the optimal hypersphere center point,

Represents the hyperparameters of the SVDD model;

Judging whether s(x') is greater than the radius of the hypersphere with the smallest volume, when s(x') is greater than the radius of the optimal hypersphere, the test point is an unsuitable area;

When s(x') is less than or equal to the radius of the optimal hypersphere, the test point is a suitable area;

The above operation is performed on all the test points of the Chinese herbal medicine to be predicted to obtain all the suitable areas of the Chinese herbal medicine to be predicted.

As shown in Figure 2, a prediction system based on the deep SVDD model for the suitable area of Chinese medicinal materials includes:

The collection module is used to collect ecological factor data of Chinese medicinal materials and generate pseudo-absence point sample data of Chinese medicinal materials;

The pseudo-absence point sample data is the unsuitable area of Chinese medicinal materials obtained by the MaxEnt model;

The preprocessing module is used for preprocessing the collected ecological factor data of Chinese medicinal materials to obtain ecological factor preprocessing data;

The prediction model generation module is used for constructing the prediction model of the suitable area for Chinese medicinal materials according to the ecological factor preprocessing data, the pseudo-absence point sample data and the SVDD model;

The prediction module is used to predict and obtain the suitable area of the Chinese herbal medicine to be predicted.

Further, the prediction process of the prediction module: select the test point of the Chinese medicinal material to be predicted in any of the ecological factor data;

Calculate the distance between the test point of the Chinese medicinal material to be predicted and the optimal hypersphere center point in the prediction model of the Chinese medicinal material suitable area:

表示SVDD模型的超参数；Among them, x' represents the test point, s(x') represents the distance between the test point and the center point of the hypersphere where the prediction model of the suitable area for Chinese medicinal materials is optimal,

Represents the hyperparameters of the SVDD model;

Judging whether s(x') is greater than the radius of the hypersphere with the smallest volume, when s(x') is greater than the radius of the optimal hypersphere, the test point is an unsuitable area;

When s(x') is less than or equal to the radius of the optimal hypersphere, the test point is a suitable area; perform the above operation on all the test points of the Chinese herbal medicine to be predicted to obtain all suitable conditions of the Chinese herbal medicine to be predicted. living area.

Example 2

As shown in Figure 3, on the basis of Example 1, along with the rising demand of Salvia miltiorrhiza, the present invention takes Salvia miltiorrhiza as the research object, and obtains a total of 120 pieces of data on the distribution of samples of Salvia miltiorrhiza; a total of 26 environmental factors are selected, such as Table 1 shows the ecological environment factors and distribution list of Chinese medicinal materials, including: 19 climate factors, 3 terrain factors, and 4 soil factors; there are 120 pseudo-absence point sample data.

240 samples of Salvia miltiorrhiza were used to verify the validity of the model. The training set and test set accounted for 80% and 20% respectively; the learning rate was set to 0.0001; On the basis of operation, all training sets are fully trained in the entire network; the batch sample size is set to 20, and the weight decay coefficient is set to 5e-07.

Using the AUC value as the evaluation index, the AUC value of this embodiment is obtained as 0.997, and the AUC value of the MaxEnt model is 0.899.

Table 1 List of ecological environment factors and distribution of Chinese medicinal materials

The specific embodiments described above further describe the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above descriptions are only specific embodiments of the present invention, and are not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (7)

1. a Chinese medicinal material suitable area prediction method based on deep SVDD model, is characterized in that, comprises: S1: Collect ecological factor data of Chinese herbal medicines, and use MaxEnt model to generate pseudo-absence point sample data of Chinese herbal medicines; The pseudo-absence point sample data is the unsuitable area of Chinese medicinal materials obtained by the MaxEnt model; S2: preprocessing the collected ecological factor data of Chinese medicinal materials to obtain ecological factor preprocessing data; S3: constructing a prediction model for the suitable area for Chinese medicinal materials according to the ecological factor preprocessing data, the pseudo-absence point sample data and the SVDD model; The construction process of the prediction model of the suitable area for Chinese medicinal materials in S3: The SVDD model uses a fully connected network

mapping the ecological factor preprocessing data to a high-dimensional feature space; Find the optimal hypersphere in the high-dimensional feature space, and the pseudo-absence point sample data is located outside the hypersphere of the optimal hypersphere, and the sample distribution data in the ecological factor data is located in the the interior of the optimal hypersphere; The formula for the radius and center of the optimal hypersphere:

where c represents the center point, and

n is the sample size,

denote each connected network, α _i and α _j are Lagrangian multipliers,

represents the inner product,

and

both represent inner product,

represents the support vector, and

S4: Put the test point of the Chinese herbal medicine to be predicted into the prediction model of the Chinese herbal medicine suitable area for judgment, and obtain the suitable area of the Chinese herbal medicine to be predicted. 2 . The method for predicting suitable areas for Chinese medicinal materials based on a deep SVDD model according to claim 1 , wherein the ecological factor data includes sample distribution data, environmental factor data and map data. 3 . 3. The method for predicting the suitable area for Chinese medicinal materials based on the deep SVDD model according to claim 1, wherein the step of generating the pseudo-absent point sample data in the S1 comprises: Use MaxEnt model to generate the numerical results of the suitable area of Chinese medicinal materials; Eliminate the values that are greater than or equal to the threshold value in the result of the suitable area of the Chinese medicinal material to obtain the non-suitable area; The same number of pseudo-absence points as the ecological factor data of Chinese medicinal materials are selected from the non-suitable area to obtain sample data of pseudo-absence points of traditional Chinese medicinal materials. 4. The method for predicting the suitable growth area of Chinese medicinal materials based on the deep SVDD model according to claim 1, wherein the preprocessing process of the S2 comprises: The word vector model Word2vec is used to convert the ecological factor data into a high-dimensional space word vector; Use the t-SNE algorithm to map high-dimensional spatial word vectors to two-dimensional spatial word vectors. 5. a kind of Chinese medicinal material suitable area prediction method based on deep SVDD model according to claim 4, is characterized in that, described t-SNE algorithm: Construct a probability distribution between high-dimensional objects, and the similarity between different data is expressed as:

Among them, p _j|i represents the similarity between different data in the high-dimensional space, x _i and x _j are any two different data in the N-dimensional data x ₁ , x ₂ ,..., x _N , The parameter σ _i represents the variance of the Gaussian distribution centered on x _i , || || represents the two-norm operation, x _k represents the N-dimensional data x ₁ , x ₂ ,..., the data subscript k in x _N ; The probability distribution is constructed for the high-dimensional objects in the low-dimensional space, and the similarity between different data is expressed as:

Among them, q _j|i represents the similarity between different data in the low-dimensional space, y _i and y _j represent the two-dimensional data y ₁ , y ₂ , y _k in the low-dimensional space, and the subscript k in the low-dimensional space two-dimensional data; Construct the joint probability distributions P and Q of the high-dimensional space and the low-dimensional space respectively, so that for any i and j, there are q _i|j =p _j|i , q _i|j =q _j|i ;

Among them, pi _{, j} represent the joint probability between any two data in the high-dimensional space, q _{i, j} represent the joint probability between any two data in the low-dimensional space, y _l represents the subscript l in the low-dimensional space two-dimensional data; Using KL divergence to measure the similarity of the joint probability distribution of high-dimensional space and low-dimensional space, we get:

Among them, C represents the similarity of the joint probability distribution of high-dimensional space and low-dimensional space, P represents the joint probability of high-dimensional space, and Q represents the joint probability of low-dimensional space. 6. A method for predicting suitable areas for Chinese medicinal materials based on a deep SVDD model according to claim 1, wherein in said S4, a test point of the Chinese medicinal material to be predicted in any said ecological factor data is selected; Calculate the distance between the test point of the Chinese medicinal material to be predicted and the center point of the optimal hypersphere:

Among them, x' represents the test point, s(x') represents the distance between the test point and the optimal hypersphere center point,

Represents the hyperparameters of the SVDD model; Judging whether s(x') is greater than the radius of the hypersphere with the smallest volume, when s(x') is greater than the radius of the optimal hypersphere, the test point is an unsuitable area; When s(x') is less than or equal to the radius of the optimal hypersphere, the test point is a suitable area; The above operation is performed on all the test points of the Chinese herbal medicine to be predicted to obtain all the suitable areas of the Chinese herbal medicine to be predicted. 7. A Chinese medicinal material suitable area prediction system based on deep SVDD model, is characterized in that, comprises: The collection module is used to collect ecological factor data of Chinese medicinal materials and generate pseudo-absence point sample data of Chinese medicinal materials; The pseudo-absence point sample data is the unsuitable area of Chinese medicinal materials obtained by the MaxEnt model; The preprocessing module is used for preprocessing the collected ecological factor data of Chinese medicinal materials to obtain ecological factor preprocessing data; The prediction model generation module is used for constructing the prediction model of the suitable area for Chinese medicinal materials according to the ecological factor preprocessing data, the pseudo-absence point sample data and the SVDD model; The prediction module is used to predict and obtain the suitable area of the Chinese herbal medicine to be predicted; The prediction process of the prediction module: select the test points of the Chinese medicinal materials to be predicted in any of the ecological factor data; Calculate the distance between the test point of the Chinese medicinal material to be predicted and the optimal hypersphere center point in the prediction model of the Chinese medicinal material suitable area:

Among them, x' represents the test point, s(x') represents the distance between the test point and the center point of the hypersphere where the prediction model of the suitable area for Chinese medicinal materials is optimal,

Represents the hyperparameters of the SVDD model; Judging whether s(x') is greater than the radius of the hypersphere with the smallest volume, when s(x') is greater than the radius of the optimal hypersphere, the test point is an unsuitable area; When s(x') is less than or equal to the radius of the optimal hypersphere, the test point is a suitable area; perform the above operation on all the test points of the Chinese herbal medicine to be predicted to obtain all suitable conditions of the Chinese herbal medicine to be predicted. living area.

Images