Plant growth model generation method and system based on machine learning

Abstract

The present invention relates to the field of data processing technology, and in particular to a method and system for generating a plant growth model based on machine learning. The method comprises: obtaining the values ​​of various parameters in each day during the plant growth process; dividing all data points into two subsets according to the division point; determining the degree of confusion of the target data point in the secondary target parameter after the division; calculating the adjustment factor of the target data point in the target parameter; determining the adjusted Gini impurity corresponding to the target data point in the target parameter; constructing a decision tree model according to the adjusted Gini impurity, obtaining a plant growth model, and completing the generation of a plant growth model based on machine learning. By comprehensively considering the conditions of various parameters, the Gini impurity used to construct the decision tree model is adjusted, and the situation that the decision tree model only considers the Gini impurity of one parameter is changed, so as to effectively alleviate the overfitting problem in the construction of the plant growth model and improve the accuracy of the plant growth model.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and system for generating a plant growth model based on machine learning. Background Art

[0002] With the continuous reduction of agricultural land, traditional agricultural production methods can no longer meet the needs of efficient and refined management of modern agriculture. Therefore, how to achieve sustainable agriculture on limited land and resources has become a hot topic in research and practice. In this context, machine learning-based plant growth model generation methods and systems are becoming an indispensable part of modern agricultural development, promoting the development of intelligent and precise agriculture, and making important contributions to achieving food security and ecological balance. How to more accurately construct plant growth models and evaluate plant growth trends has become an urgent problem to be solved.

[0003] The patent document with publication number CN111309852B discloses a method, system, device and storage medium for generating a visualized decision tree (set) model, wherein the method includes: organizing the decision knowledge in the medical guide into a rule set in a data structured form; splitting the rule set, and counting the number of decision trees N, N≥1, without supplementing the missing values; taking the rule sets corresponding to the N decision trees respectively, constructing a decision tree model with the rule set as input and a decision tree as output, and generating a data structure model for each decision tree; visualizing the data structure model of each constructed decision tree to generate a visualized decision tree model, and saving the visualized decision tree model in the form of an image.

[0004] However, the above patent does not solve the problem that when generating a decision tree model, during the process of dividing the plant growth data, the selection of each division point only considers the Gini impurity of one parameter at the division point, without comprehensive consideration, resulting in a division for each parameter with the smallest Gini impurity. As a result, the construction of the model is prone to overfitting. Overfitting means that the model performs better on the training data, but has worse subsequent prediction accuracy, making the model unable to effectively respond to changes in different environments, different crop growth conditions, etc., resulting in a significant reduction in the model's predictive ability, and being unable to reflect the growth status of plants under various conditions, reducing the model's accuracy. At the same time, overfitting will complicate the model, increase the difficulty of understanding for plant managers, and increase the burden of calculation and storage. Summary of the invention

[0005] In order to solve the problem that during the division process of the decision tree model, the selection of each division point only considers the Gini impurity of one parameter of the division point, resulting in a division for the minimum Gini impurity of each parameter, which makes the construction of the decision tree model more prone to overfitting, thereby reducing the accuracy of the model, increasing the complexity of the model, and increasing the calculation and storage burden, the present invention provides a plant growth model generation method and system based on machine learning.

[0006] In a first aspect, the present invention provides a method for generating a plant growth model based on machine learning, which adopts the following technical solution:

[0007] The method for generating a plant growth model based on machine learning includes: obtaining the values ​​of various parameters every day during the plant growth process; taking each day as a data point, recording any data point as a target data point, recording any parameter as a target parameter, taking the target data point as a dividing point, and dividing all data points into two subsets according to the values ​​of all data points and the target data point in the target parameter; recording any parameter of the remaining parameters not including the target parameter as a sub-target parameter, calculating the average value of the data points in the two subsets in the sub-target parameter, and calculating the minimum value of the data points in the subset with the largest average value, as well as the average value. The maximum value of the data point in the smallest subset is used to determine the degree of confusion of the sub-target parameter after the target data point is used as the dividing point; the weights of each parameter are obtained by using the hierarchical analysis method, and the adjustment factor of the target data point in the target parameter is calculated based on the weight and the degree of confusion; the Gini impurity corresponding to the target data point in the target parameter for building a decision tree model is weighted with the adjustment factor as the weight to obtain the adjusted Gini impurity corresponding to the target data point in the target parameter; according to the adjusted Gini impurity, a decision tree model is constructed to obtain a plant growth model, thereby completing the generation of a plant growth model based on machine learning.

[0008] By adjusting the Gini impurity, the decision tree model's excessive reliance on a single parameter is reduced, thereby effectively avoiding overfitting and improving the generalization ability of the model; the introduction of the hierarchical analysis method and the degree of confusion calculation ensures that appropriate features can be scientifically selected during the construction of the decision tree model, thereby improving the accuracy of the model; by weighted adjustment of the Gini impurity and considering the mutual influence of multiple parameters, the division is made more refined and accurate, thereby improving the prediction performance of the model; the data processing process is optimized, unnecessary complexity is reduced, and the model calculation is made more efficient; by dynamically adjusting the parameter weights, the model can adapt to the changes of different plants and growth environments and is more applicable.

[0009] Furthermore, the various parameters include plant height, stem thickness, number of leaves, and average grayscale value of the collected plant image.

[0010] Furthermore, the two subsets are respectively a first subset and a second subset; wherein the first subset is a set of data points whose values ​​in the target parameter are greater than the values ​​of the target data points in the target parameter; and the second subset is a set of data points whose values ​​in the target parameter are not greater than the values ​​of the target data points in the target parameter.

[0011] Furthermore, the degree of confusion satisfies the following relationship:

[0012] ;

[0013] In the formula, For the first The data point is the remaining parameters after the division point, excluding the target parameter. The degree of confusion of the item parameters, is the maximum value of the data point in the subset with the smallest mean value, is the minimum value of the data point in the subset with the largest mean value, For the The range of the data points in the item parameter, For the The value in the item parameter is and The number of data points between For the The number of data points in the item parameter.

[0014] By optimizing the division of data points, the confusion between categories is reduced, and the accuracy and generalization ability of the model are improved; by weighting the range, maximum and minimum values ​​and other features of the data points, the data division is improved, making the data easier to distinguish in high-dimensional or complex situations; by identifying and excluding data points that have little impact on classification, the noise is reduced and the robustness of the model is improved; the division strategy is dynamically adjusted according to the data distribution, so that different types of data can be processed in a more adaptive way.

[0015] Furthermore, the adjustment factor satisfies the following relationship:

[0016] ;

[0017] In the formula, For the The data point in The adjustment factor in the item parameter, is the total number of parameters, For the The weight of the item parameter, The remaining parameters excluding the target parameter The weight of the item parameter, After the division The data point is in the remaining parameters excluding the target parameter. The degree of obfuscation of the item parameters.

[0018] By adjusting the factors, the model can dynamically optimize the weight distribution of each parameter according to the relationship between the data distribution and the parameters, thereby improving the performance of the model; the adjustment factors can fine-tune the influence of each data point on different parameters, reduce errors, and improve classification or regression accuracy; by comprehensively considering the degree of confusion and parameter weights, the overfitting of specific data points or parameters can be reduced, thereby enhancing the robustness of the model.

[0019] Furthermore, the adjusted Gini impurity satisfies the following relationship:

[0020] ;

[0021] In the formula, For the The data point in The corresponding adjusted Gini impurity in the term parameter, For the The data point in The Gini impurity corresponding to the item parameter used to build the decision tree model, For the The data point in The adjustment factor in the item parameter.

[0022] After the introduction of the adjustment factor, Gini impurity can more accurately reflect the classification difficulty of each data point, helping the decision tree to make more discriminative decisions at each split; the adjustment factor helps reduce the impact of noise data, allowing the decision tree model to focus on valuable features, thereby reducing the risk of overfitting and improving the generalization ability of the model. By dynamically adjusting the Gini impurity, the model can automatically adapt to changes in different features and data distribution, allowing the decision tree to learn and predict more effectively.

[0023] Furthermore, the plant growth model includes a plurality of divided branches, and a characteristic sequence of each branch is obtained. The plant growth stage corresponding to each branch is determined according to a characteristic standard, and the characteristic sequence is a sequence composed of the median values ​​of various parameters in the branch.

[0024] Using the median as the feature representative reduces the impact of extreme data and noise, making the model more robust and thus improving the accuracy of judging the plant growth stage; the median has good adaptability to asymmetric data distribution, reduces the risk of overfitting, and enables the model to better adapt to new data.

[0025] In a second aspect, the present invention provides a plant growth model generation system based on machine learning, which adopts the following technical solutions:

[0026] A plant growth model generation system based on machine learning includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the plant growth model generation method based on machine learning is implemented.

[0027] By adopting the above technical solution, the above-mentioned plant growth model generation method based on machine learning is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0028] The present invention has the following technical effects:

[0029] By comprehensively considering various parameters, the Gini impurity used to construct the decision tree model is adjusted, which changes the situation in which the original decision tree model only considers one parameter, the Gini impurity, and effectively alleviates the overfitting problem that is prone to occur in the construction of the plant growth model, thereby improving the accuracy of the plant growth model; avoiding the situation in which the model complexity is too high due to overfitting, reducing unnecessary complex calculations, reducing the computational and storage burden, and making the model more feasible and efficient in practical applications; using the hierarchical analysis method to obtain parameter weights, and combining the relevant confusion degree to adjust the Gini impurity, thereby constructing a plant growth model that is more in line with the actual plant growth situation and more reasonable and accurate, which helps to better analyze and predict the plant growth process. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding numbers are the same or corresponding parts.

[0031] Figure 1 It is a flow chart of a method for generating a plant growth model based on machine learning in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0033] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0034] In the process of dividing various parameters in the plant growth process, the present invention divides the decision tree according to the Gini impurity of only one parameter in the conventional decision tree model, and in the process of recursive division, each parameter is divided once according to the minimum Gini impurity of each parameter, resulting in overfitting in the construction of the model. Therefore, in each division process, the parameters of the two subsets after the division point is divided in the target parameter, the confusion degree of the remaining parameters is obtained, and the adjustment factor of the target data point in the target parameter is weighted to obtain a more accurate adjusted Gini impurity to construct the decision tree model, so as to obtain a plant growth model.

[0035] The present invention discloses a method for generating a plant growth model based on machine learning. Figure 1 , comprising steps S1 to S6:

[0036] S1: Obtain the values ​​of various parameters every day during the plant growth process.

[0037] Specifically, the parameters include plant height, stem thickness, number of leaves, and average grayscale value of the collected plant image.

[0038] By using an agricultural camera to take pictures of plants at fixed points every day to obtain original images, semantic segmentation is used to obtain plant images in the original images, and then the plant height, stem thickness, number of leaves, and average grayscale value of the collected plant images in this embodiment are obtained. For different types of plants, implementers can make adjustments based on specific implementation circumstances.

[0039] S2: Divide all data points into two subsets according to the partition point.

[0040] Take each day as a data point, record any data point as the target data point, record any parameter as the target parameter, take the target data point as the dividing point, and divide all data points into two subsets according to the values ​​of all data points and the target data point in the target parameter.

[0041] Specifically, the two subsets are the first subset and the second subset respectively; wherein the first subset is a set of data points whose values ​​in the target parameter are greater than the values ​​of the target data points in the target parameter; and the second subset is a set of data points whose values ​​in the target parameter are not greater than the values ​​of the target data points in the target parameter.

[0042] S3: Determine the degree of confusion of the target data points in the secondary target parameters after division.

[0043] It should be noted that the construction process of the decision tree model is achieved by continuously dividing the data set. At each division, it is necessary to evaluate the quality of the division, that is, whether the subset obtained after the division is more "pure", that is, whether the values ​​of the data points in the subset on each feature are more consistent. By calculating the degree of confusion of the secondary target parameter, you can intuitively understand the degree of mixing of the two subsets on this parameter after the division. The lower the degree of confusion, the purer the subset is on this parameter, and the better the division effect. When building a decision tree model, it is necessary to select the best division point from many possible division points. By calculating the degree of confusion when different data points are used as division points, the division point with the minimum degree of confusion can be found, thereby determining the best division scheme, so that the subset after division has the minimum confusion on the secondary target parameter, and improving the decision tree model's ability to classify data.

[0044] Any parameter among the remaining parameters that do not contain the target parameter is recorded as the secondary target parameter, and the average value of the data points in the two subsets in the secondary target parameter is calculated. According to the minimum value of the data point in the subset with the largest average value and the maximum value of the data point in the subset with the smallest average value, the degree of confusion of the secondary target parameter after the target data point is used as the dividing point is determined. If the calculated average values ​​of the data points in the two subsets in the secondary target parameter are equal, the median value in any subset is selected and recorded as , the median value in the other subset is recorded as , to complete subsequent calculations.

[0045] The degree of confusion satisfies the following relationship:

[0046] ;

[0047] In the formula, For the first The data point is the remaining parameters after the division point, excluding the target parameter. The degree of confusion of the item parameters, is the maximum value of the data point in the subset with the smallest mean value, is the minimum value of the data point in the subset with the largest mean value, For the The range of the data points in the item parameter, For the The value in the item parameter is and The number of data points between For the The number of data points in the item parameter.

[0048] in, When , it indicates that there is a confusion area in the secondary target parameter, that is, the repeated area of ​​the data range of the two subsets. At this time, the larger the confusion area, the greater the degree of confusion in the secondary target parameter. When , it means that there is no obvious cross-repetitive area in the sub-target parameters of the two subsets, and the degree of confusion in the sub-target parameters is smaller; Indicates the degree of overlap between the two subsets in the secondary target parameter. The larger the value, the greater the overlap. Indicates The range of the data points in the parameter. The larger the value, the The larger the data range of the item parameter, the The smaller the proportion of the entire value range, the less confusion there is; Indicates The data points where the parameter is located at the junction of the two subsets account for The ratio of the total number of data points in the parameter. The larger the ratio, the better. When the target parameter is divided, the more data points there are between the two divided subsets, the more data points are in a confused state and are not clearly separated.

[0049] S4: Calculate the adjustment factor of the target data point in the target parameter.

[0050] It should be noted that when any data point is used as a partition point, the greater the corresponding confusion level, the less ideal the partitioning effect of the partition point on the data, and the preference for selecting such a partition point should be lower. In the decision tree model, since the minimum Gini impurity is selected, in order to more reasonably measure the pros and cons of each data point as a partition point in the target parameter, it is necessary to comprehensively consider the importance of different parameters and the degree of confusion generated after the partition, and calculate the adjustment factor to facilitate the subsequent adjustment of the Gini impurity.

[0051] The weights of various parameters are obtained by using the hierarchical analysis method, and the adjustment factor of the target data point in the target parameter is calculated based on the weights and the confusion degree.

[0052] Specifically, the adjustment factor satisfies the following relationship:

[0053] ;

[0054] In the formula, For the The data point in The adjustment factor in the item parameter, is the total number of parameters, For the The weight of the item parameter, The remaining parameters excluding the target parameter The weight of the item parameter, After the division The data point is in the remaining parameters excluding the target parameter. The degree of obfuscation of the item parameters.

[0055] in, represents the weight importance coefficient, The larger the The more important the item parameter is, the smaller the adjustment factor is; Indicates the When the data points are divided into The degree of confusion of the remaining parameters other than the item parameter is weighted by the corresponding weights. The larger the value, the more confused the performance of the two subsets after division in the remaining parameters, so the larger the adjustment factor.

[0056] S5: Determine the adjusted Gini impurity corresponding to the target data point in the target parameters.

[0057] It should be noted that the Gini impurity is adjusted by adjusting the adjustment factor so that data points with a higher degree of confusion after division have a higher adjusted Gini impurity, making it more difficult to be selected for data division during the decision tree division process.

[0058] The adjustment factor is used as a weight to weight the Gini impurity corresponding to the target data point in the target parameter for building the decision tree model, so as to obtain the adjusted Gini impurity corresponding to the target data point in the target parameter.

[0059] Specifically, the adjusted Gini impurity satisfies the following relationship:

[0060] ;

[0061] In the formula, For the The data point in The corresponding adjusted Gini impurity in the term parameter, For the The data point in The Gini impurity corresponding to the item parameter used to build the decision tree model, For the The data point in The adjustment factor in the item parameter.

[0062] S6: According to the adjusted Gini impurity, a decision tree model is constructed to obtain a plant growth model, thereby completing the generation of a plant growth model based on machine learning.

[0063] Specifically, the plant growth model includes a plurality of divided branches, a characteristic sequence of each branch is obtained, and the plant growth stage corresponding to each branch is determined according to a characteristic standard, wherein the characteristic sequence is a sequence composed of the median values ​​of various parameters in the branch.

[0064] An embodiment of the present invention also discloses a plant growth model generation system based on machine learning, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a plant growth model generation method based on machine learning according to the present invention is implemented.

[0065] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0066] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, or any other medium that can be used to store the required information and can be accessed by an application, module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0067] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0068] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating a plant growth model based on machine learning, characterized in that: include: Obtaining the values ​​of various parameters every day during the growth of the plant, wherein the various parameters include the plant height, stem thickness, number of leaves, and average grayscale value of the collected plant image; Take each day as a data point, record any data point as the target data point, record any parameter as the target parameter, take the target data point as the dividing point, and divide all data points into two subsets according to the values ​​of all data points and the target data point in the target parameter; record any parameter among the remaining parameters that do not contain the target parameter as the sub-target parameter, calculate the average value of the data points in the two subsets in the sub-target parameter, and determine the degree of confusion of the sub-target parameter after taking the target data point as the dividing point according to the minimum value of the data point in the subset with the largest average value and the maximum value of the data point in the subset with the smallest average value; The weights of various parameters are obtained by using the hierarchical analysis method. Based on the weights and the degree of confusion, the adjustment factor of the target data point in the target parameter is calculated. The adjustment factor satisfies the following relationship: ; In the formula, For the The data point in The adjustment factor in the item parameter, is the total number of parameters, For the The weight of the item parameter, The remaining parameters excluding the target parameter The weight of the item parameter, After the division The data point is in the remaining parameters excluding the target parameter. The degree of confusion of item parameters; Using the adjustment factor as a weight, weighting the Gini impurity corresponding to the target data point in the target parameter for building the decision tree model, to obtain the adjusted Gini impurity corresponding to the target data point in the target parameter; According to the adjusted Gini impurity, a decision tree model is constructed to obtain a plant growth model, thereby completing the generation of a plant growth model based on machine learning.

2. The method for generating a plant growth model based on machine learning according to claim 1, characterized in that: The two subsets are respectively the first subset and the second subset; wherein the first subset is a set of data points whose values ​​in the target parameter are greater than the values ​​of the target data points in the target parameter; and the second subset is a set of data points whose values ​​in the target parameter are not greater than the values ​​of the target data points in the target parameter.

3. The method for generating a plant growth model based on machine learning according to claim 1, characterized in that: The degree of confusion satisfies the following relationship: ； In the formula, is the maximum value of the data point in the subset with the smallest mean value, is the minimum value of the data point in the subset with the largest mean value, For the The range of the data points in the item parameter, For the The value in the item parameter is and The number of data points between For the The number of data points in the item parameter.

4. The method for generating a plant growth model based on machine learning according to claim 1, characterized in that: The adjusted Gini impurity satisfies the following relationship: ； In the formula, For the The data point in The corresponding adjusted Gini impurity in the term parameter, For the The data point in The Gini impurity corresponding to the item parameter used to build the decision tree model.

5. The method for generating a plant growth model based on machine learning according to claim 1, characterized in that: The plant growth model includes a plurality of divided branches, and a characteristic sequence of each branch is obtained. The plant growth stage corresponding to each branch is determined according to a characteristic standard. The characteristic sequence is a sequence composed of the median values ​​of various parameters in the branch.

6. A plant growth model generation system based on machine learning, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for generating a plant growth model based on machine learning according to any one of claims 1 to 5 is implemented.

Images