A plant growth prediction method, device and equipment based on machine learning model

Abstract

The present invention relates to the field of plant prediction technology, and specifically to a plant growth prediction method, device and equipment based on a machine learning model, including: obtaining environmental data and state data of a target plant; classifying the state of the target plant according to the state data based on a trained state classification model to obtain a state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state. The present invention can more accurately reflect the growth of plants under different environmental conditions and improve the accuracy of prediction by using different prediction models for plants in normal and abnormal states, and using a trained state classification model can help quickly judge the health status of the plant, so as to take corresponding measures in time to avoid potential losses.

Description

Technical Field

[0001] The present invention relates to the technical field of plant prediction, and in particular to a plant growth prediction method, device and equipment based on a machine learning model. Background Art

[0002] Machine learning models are mathematical structures that learn from data and make predictions or decisions through algorithms. They use input data to identify patterns and generate output labels or predictions.

[0003] Traditional methods often rely on simple linear models or empirical rules, which make it difficult to capture the complex nonlinear relationship between plant growth and environmental factors, resulting in low accuracy of prediction results. Without feature selection, traditional methods may use a large number of redundant and irrelevant features, which not only increases the computational complexity, but may also affect the performance and interpretability of the model. Moreover, traditional methods usually use the same prediction model for normal and abnormal conditions and cannot effectively distinguish the needs of plants under different conditions, resulting in untimely and inaccurate decisions under abnormal conditions. Traditional methods often rely on empirical judgments and lack in-depth mining and utilization of large amounts of data, which cannot fully tap the potential of the data, leading to increased uncertainty in the decision-making process. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose a plant growth prediction method, device and equipment based on a machine learning model.

[0005] The technical solution of the present invention is a plant growth prediction method based on a machine learning model, comprising:

[0006] Obtain environmental data and status data of target plants;

[0007] Based on the trained state classification model, the state of the target plant is classified according to the state data to obtain the state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state;

[0008] Based on the trained first prediction model, growth prediction is performed on the target plant whose state type is a normal state according to the environmental data to obtain first growth data corresponding to the target plant;

[0009] Based on the trained second prediction model, growth prediction is performed on the target plant whose state type is an abnormal state according to the environmental data to obtain second growth data corresponding to the target plant.

[0010] Preferably, the status data of the target plant includes leaf area, leaf color, trunk length and trunk area.

[0011] Preferably, the state classification model adopts an improved support vector machine, and the support vector machine is improved based on an optimization algorithm to obtain the state classification model.

[0012] Preferably, the support vector machine is improved based on an optimization algorithm to obtain the state classification model, including:

[0013] Encoding various combinations of the penalty parameter and the kernel function of the support vector machine to obtain an individual population corresponding to the optimization algorithm, wherein each individual in the individual population corresponds to a combination of the penalty parameter and the kernel function;

[0014] The position of each individual in the individual population is randomly generated, wherein the position generation formula of the individual is as follows:

[0015] x i =rand(1,d)×(b u -b1)+b1;

[0016] Among them, x i represents the position of the i-th individual in the individual population, rand() represents the random generation function, d represents the total dimension of the individual population, b u represents the maximum value of the individual's position, and b1 represents the minimum value of the individual's position;

[0017] The fitness of each individual in the individual population is calculated based on the objective function of the support vector machine, and the global optimal position is obtained based on the fitness of each individual in the individual population.

[0018] Preferably, the support vector machine is improved based on an optimization algorithm to obtain the state classification model, and further includes:

[0019] The position of each individual in the individual population is updated, wherein the position update formula of the individual is as follows:

[0020] x i，j (t+1)=x i，j (t)+0.5[(2CZP pos (j)-x i，j (t))+2(1-C)Zu (j) +x i，j (t)];

[0021] Among them, x i，j (t+1) represents the position of the jth dimension of the i-th individual in the individual population at the t+1th iteration, x i，j (t) represents the position of the jth dimension of the ith individual in the individual population at the tth iteration, C represents the equilibrium parameter, and tmax represents the preset maximum number of iterations, Z represents the fitness parameter, P pos(j) represents the position of the individual, u (j) represents the average value of all positions;

[0022] The fitness of each individual in the updated individual population is calculated based on the objective function of the support vector machine, and the above steps are repeated until a preset maximum number of iterations is reached.

[0023] Preferably, obtaining environmental data of the target plant includes:

[0024] Based on the sensor group, the temperature and humidity above the canopy, soil heat flux, saturation, water vapor pressure, net solar radiation, wind speed, wind direction and soil temperature of the environment where the target plant is located are obtained.

[0025] Preferably, the first prediction model and the second prediction model both adopt an improved decision tree algorithm, and the first prediction model and the second prediction model are obtained by performing feature selection on the environmental data.

[0026] Preferably, performing feature selection on the environmental data to obtain the first prediction model and the second prediction model includes:

[0027] Randomly select a sample from the training sample set, and find multiple neighboring samples of the sample from the training sample set of the same type as the sample;

[0028] Repeat the above operation until all types of the training sample sets are traversed to obtain multiple neighbor samples corresponding to the various types of training samples;

[0029] The weights corresponding to the plurality of neighboring samples corresponding to the various types of the training samples are calculated, and feature selection is performed on the environmental data based on the weights. The weight calculation formula is as follows:

[0030]

[0031] Where W(A) represents the weights of multiple neighbor samples corresponding to various types of training samples, k represents the number of neighbor samples, class(R) represents the category corresponding to the neighbor sample R, p(C) represents the prior probability of category C, m represents the pre-set number of iterations, M j (C) represents the jth nearest neighbor sample of category C, p(Class(R)) represents the prior probability of the category corresponding to the nearest neighbor sample R, diff(A,R1,R2) represents the difference in feature A between the nearest neighbor sample R1 and the nearest neighbor sample R2, and R1(A) represents the feature A of the neighboring sample R1, H jrepresents the jth nearest neighbor sample.

[0032] The technical solution of the present invention is a plant growth prediction device based on a machine learning model, which is applicable to the plant growth prediction method based on a machine learning model, comprising:

[0033] A data acquisition unit, the data acquisition unit is used to acquire environmental data and status data of the target plant;

[0034] A state classification unit, the state classification unit is used to classify the state of the target plant according to the state data based on a trained state classification model to obtain a state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state;

[0035] a first prediction unit, the first prediction unit being used to perform growth prediction on the target plant whose state type is a normal state according to the environmental data based on a trained first prediction model, so as to obtain first growth data corresponding to the target plant;

[0036] The second prediction unit is used to perform growth prediction on the target plant whose state type is an abnormal state according to the environmental data based on the trained second prediction model to obtain second growth data corresponding to the target plant.

[0037] The technical solution of the present invention: a plant growth prediction device based on a machine learning model, which is applicable to a plant growth prediction method based on a machine learning model, comprising: at least one processor and a memory;

[0038] The memory is used to store computer-executable instructions;

[0039] The at least one processor is used to execute the computer-executable instructions stored in the memory to implement a plant growth prediction method based on a machine learning model.

[0040] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0041] 1. The present invention uses different prediction models for plants in normal and abnormal states respectively, which can more accurately reflect the growth conditions of plants under different environmental conditions and improve the accuracy of prediction. The use of a trained state classification model helps to quickly judge the health status of plants, so that corresponding measures can be taken in time to avoid potential losses. Moreover, by making separate predictions for plants in abnormal states, it can help to discover problems in time, implement intervention measures, reduce risks, and protect plant health. In addition, by using machine learning technology to model plant growth, the decision-making process is more dependent on data and models, reducing the uncertainty caused by empirical judgment.

[0042] 2. The present invention helps to remove redundant and irrelevant features and reduce computational complexity by performing feature selection on environmental data, while improving the interpretability and predictive performance of the model. The improved support vector machine can better adapt to complex data distribution and improve the accuracy of state classification, thereby more effectively distinguishing normal and abnormal states. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic flow chart of the steps of an overall method in an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of the overall structure of an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of the architecture of an overall device in an embodiment of the present invention.

[0046] Figure numerals: 1. Data acquisition unit; 2. State classification unit; 3. First prediction unit; 4. Second prediction unit; 5. Processor; 6. Memory; 7. Bus. DETAILED DESCRIPTION

[0047] Embodiment 1, as Figure 1 As shown, the present invention proposes a plant growth prediction method based on a machine learning model, comprising:

[0048] S1. Obtain environmental data and status data of target plants;

[0049] S2. Classifying the state of the target plant according to the state data based on the trained state classification model to obtain the state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state;

[0050] S3, based on the trained first prediction model, performing growth prediction on the target plant whose state type is normal according to the environmental data to obtain first growth data corresponding to the target plant;

[0051] S4. Based on the trained second prediction model, growth prediction is performed on the target plant whose state type is an abnormal state according to the environmental data to obtain second growth data corresponding to the target plant.

[0052] Embodiment 2: A plant growth prediction method based on a machine learning model proposed in the present invention, compared with Embodiment 1, this embodiment also includes: the status data of the target plant includes leaf area, leaf color, trunk length and trunk area.

[0053] In this embodiment, the leaf area is estimated based on the length and width of the leaf by manually measuring the length and width of the leaf; the leaf color is directly compared and measured by using a color card or a chromatograph to record its color characteristics; the trunk length is physically measured from the root to the top by using a tape measure or a ruler; the trunk area is estimated based on the diameter or circumference of the trunk by manually measuring the diameter or circumference of the trunk.

[0054] In an optional embodiment, the state classification model adopts an improved support vector machine, and the support vector machine is improved based on an optimization algorithm to obtain the state classification model.

[0055] It should be noted that support vector machine is a supervised learning model, mainly used for classification and regression analysis. It separates data of different categories by finding an optimal hyperplane. The optimization algorithm is used to find the optimal solution of a certain objective function. Many model training processes in machine learning involve the optimization of the loss function.

[0056] In an optional embodiment, the support vector machine is improved based on an optimization algorithm to obtain a state classification model, including:

[0057] Encoding various combinations of penalty parameters and kernel functions of the support vector machine to obtain an individual population corresponding to the optimization algorithm, wherein each individual in the individual population corresponds to a combination of the penalty parameter and the kernel function;

[0058] The position of each individual in the individual population is randomly generated, where the individual position generation formula is as follows:

[0059] x i =rand(1,d)×(b u -b1)+b1;

[0060] Among them, x i represents the position of the i-th individual in the individual population, rand() represents the random generation function, d represents the total dimension of the individual population, b u represents the maximum value of the individual's position, and b1 represents the minimum value of the individual's position;

[0061] The fitness of each individual in the individual population is calculated based on the objective function of the support vector machine, and the global optimal position is obtained based on the fitness of each individual in the individual population.

[0062] In an optional embodiment, the support vector machine is improved based on an optimization algorithm to obtain a state classification model, further comprising:

[0063] The position of each individual in the individual population is updated, where the individual position update formula is as follows:

[0064] x i,j (t+1)=x i,j (t)+0.5[(2CZP pos(j) -x i,j (t))+2(1-C)Zu (j) +x i,j (t)];

[0065] Among them, x i,j (t+1) represents the position of the jth dimension of the i-th individual in the individual population at the t+1th iteration, x i,j (t) represents the position of the jth dimension of the ith individual in the individual population at the tth iteration, C represents the equilibrium parameter, and t max represents the preset maximum number of iterations, Z represents the fitness parameter, P pos(j) represents the position of the individual, u (j) represents the average value of all positions;

[0066] The fitness of each individual in the updated individual population is calculated based on the objective function of the support vector machine, and the above steps are repeated until the preset maximum number of iterations is reached.

[0067] In an optional embodiment, obtaining environmental data of a target plant includes:

[0068] Based on the sensor group, the temperature and humidity above the canopy, soil heat flux, saturation, water vapor pressure, net solar radiation, wind speed, wind direction and soil temperature of the target plant's environment are obtained.

[0069] In an optional embodiment, both the first prediction model and the second prediction model adopt an improved decision tree algorithm, and feature selection is performed on environmental data to obtain the first prediction model and the second prediction model.

[0070] It should be noted that the decision tree algorithm is a commonly used machine learning method, mainly used for classification and regression tasks. It uses a tree structure to represent the decision-making process, which is intuitive and easy to explain. Due to the different states of plants, the feature weights required by the decision tree are also different. Therefore, the state of the plant is first classified, and then different decision trees are constructed to predict the growth of the plant.

[0071] In an optional embodiment, feature selection is performed on environmental data to obtain a first prediction model and a second prediction model, including:

[0072] Randomly select a sample from the training sample set, and find multiple neighboring samples of the sample from the same training sample set as the sample;

[0073] Repeat the above operation until all types of training sample sets are traversed to obtain multiple neighbor samples corresponding to various types of training samples;

[0074] Calculate the weights of multiple neighboring samples corresponding to various types of training samples, and perform feature selection on environmental data based on the weights. The weight calculation formula is as follows:

[0075]

[0076] Where W(A) represents the weights of multiple neighbor samples corresponding to various types of training samples, k represents the number of neighbor samples, class(R) represents the category corresponding to the neighbor sample R, p(C) represents the prior probability of category C, m represents the pre-set number of iterations, M j (C) represents the jth nearest neighbor sample of category C, p(Class(R)) represents the prior probability of the category corresponding to the nearest neighbor sample R, diff(A,R1,R2) represents the difference in feature A between the nearest neighbor sample R1 and the nearest neighbor sample R2, and R1(A) represents the feature A of the neighboring sample R1, H j represents the jth nearest neighbor sample.

[0077] Embodiment three, as Figure 2 As shown, a plant growth prediction device based on a machine learning model, which is applicable to the above-mentioned plant growth prediction method based on a machine learning model, includes:

[0078] A data acquisition unit 1, the data acquisition unit 1 is used to acquire environmental data and status data of the target plant;

[0079] A state classification unit 2, the state classification unit 2 is used to classify the state of the target plant according to the state data based on the trained state classification model to obtain the state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state;

[0080] A first prediction unit 3, the first prediction unit 3 is used to perform growth prediction on a target plant whose state type is a normal state according to environmental data based on a trained first prediction model, so as to obtain first growth data corresponding to the target plant;

[0081] The second prediction unit 4 is used to perform growth prediction on a target plant whose state type is an abnormal state according to environmental data based on a trained second prediction model to obtain second growth data corresponding to the target plant.

[0082] Embodiment 4, as Figure 3As shown, a plant growth prediction device based on a machine learning model, which is applicable to a plant growth prediction method based on a machine learning model, comprises: at least one processor 5 and a memory 6;

[0083] The memory 6 is used to store computer execution instructions;

[0084] At least one processor 5 is used to execute computer-executable instructions stored in the memory 6 to implement a plant growth prediction method based on a machine learning model.

[0085] In an optional embodiment, the memory 6 can be independent or integrated with the processor 5. When the memory 6 is independently provided, the device further includes a bus 7 for connecting the memory 6 and the processor 5.

[0086] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A plant growth prediction method based on a machine learning model, characterized in that: include: Obtain environmental data and status data of target plants; Based on the trained state classification model, the state of the target plant is classified according to the state data to obtain the state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state; Based on the trained first prediction model, growth prediction is performed on the target plant whose state type is a normal state according to the environmental data to obtain first growth data corresponding to the target plant; Based on the trained second prediction model, growth prediction is performed on the target plant whose state type is an abnormal state according to the environmental data to obtain second growth data corresponding to the target plant.

2. The plant growth prediction method based on a machine learning model according to claim 1, characterized in that: The state data of the target plant includes leaf area, leaf color, trunk length and trunk area.

3. The plant growth prediction method based on a machine learning model according to claim 1, characterized in that: The state classification model adopts an improved support vector machine, and the support vector machine is improved based on an optimization algorithm to obtain the state classification model.

4. The plant growth prediction method based on a machine learning model according to claim 3, characterized in that: The support vector machine is improved based on an optimization algorithm to obtain the state classification model, including: Encoding various combinations of the penalty parameter and the kernel function of the support vector machine to obtain an individual population corresponding to the optimization algorithm, wherein each individual in the individual population corresponds to a combination of the penalty parameter and the kernel function; The position of each individual in the individual population is randomly generated, wherein the position generation formula of the individual is as follows: x i =rand(1,d)×(b u -b1)+b1; Among them, x i represents the position of the i-th individual in the individual population, rand() represents the random generation function, d represents the total dimension of the individual population, b u represents the maximum value of the individual's position, and b1 represents the minimum value of the individual's position; The fitness of each individual in the individual population is calculated based on the objective function of the support vector machine, and the global optimal position is obtained based on the fitness of each individual in the individual population.

5. The plant growth prediction method based on a machine learning model according to claim 4, characterized in that: The support vector machine is improved based on an optimization algorithm to obtain the state classification model, and further includes: The position of each individual in the individual population is updated, wherein the position update formula of the individual is as follows: x i,j (t+1)=x i,j (t)+0.5[(2CZP pos(j) -x i,j (t))+2(1-C)To (j) +x i,j (t)]; Among them, x i,j (t+1) represents the position of the jth dimension of the i-th individual in the individual population at the t+1th iteration, x i,j (t) represents the position of the jth dimension of the ith individual in the individual population at the tth iteration, C represents the equilibrium parameter, and t max represents the preset maximum number of iterations, Z represents the fitness parameter, and P pos(j) represents the position of the individual, u (j) represents the average value of all positions; The fitness of each individual in the updated individual population is calculated based on the objective function of the support vector machine, and the above steps are repeated until a preset maximum number of iterations is reached.

6. The plant growth prediction method based on a machine learning model according to claim 1, characterized in that: Obtain environmental data of target plants, including: Based on the sensor group, the temperature and humidity above the canopy, soil heat flux, saturation, water vapor pressure, net solar radiation, wind speed, wind direction and soil temperature of the environment where the target plant is located are obtained.

7. The plant growth prediction method based on a machine learning model according to claim 1, characterized in that: The first prediction model and the second prediction model both adopt an improved decision tree algorithm, and the first prediction model and the second prediction model are obtained by performing feature selection on the environmental data.

8. The plant growth prediction method based on a machine learning model according to claim 1, characterized in that: The first prediction model and the second prediction model are obtained by performing feature selection on the environmental data, including: Randomly select a sample from the training sample set, and find multiple neighboring samples of the sample from the training sample set of the same type as the sample; Repeat the above operation until all types of the training sample sets are traversed to obtain multiple neighbor samples corresponding to the various types of the training samples; The weights corresponding to the plurality of neighboring samples corresponding to the various types of the training samples are calculated, and feature selection is performed on the environmental data based on the weights. The weight calculation formula is as follows: Where W(A) represents the weights of multiple neighbor samples corresponding to various types of training samples, k represents the number of neighbor samples, class(R) represents the category corresponding to the neighbor sample R, p(C) represents the prior probability of category C, m represents the pre-set number of iterations, M j (C) represents the jth nearest neighbor sample of category C, p(Class(R)) represents the prior probability of the category corresponding to the nearest neighbor sample R, diff(A,R1,R2) represents the difference in feature A between the nearest neighbor sample R1 and the nearest neighbor sample R2, and R1(A) represents the feature A of the neighboring sample R1, H j represents the jth nearest neighbor sample.

9. A plant growth prediction device based on a machine learning model, which is applicable to a plant growth prediction method based on a machine learning model according to any one of claims 1 to 8, characterized in that: include: A data acquisition unit (1), the data acquisition unit (1) is used to acquire environmental data and status data of a target plant; A state classification unit (2), the state classification unit (2) being used to classify the state of the target plant according to the state data based on a trained state classification model, so as to obtain a state type corresponding to the target plant, wherein the state type includes an abnormal state and a normal state; A first prediction unit (3), the first prediction unit (3) being used to perform growth prediction on the target plant whose state type is a normal state according to the environmental data based on a trained first prediction model, so as to obtain first growth data corresponding to the target plant; The second prediction unit (4) is used to perform growth prediction on the target plant whose state type is an abnormal state according to the environmental data based on the trained second prediction model, so as to obtain second growth data corresponding to the target plant.

10. A plant growth prediction device based on a machine learning model, which is applicable to a plant growth prediction method based on a machine learning model according to any one of claims 1 to 8, characterized in that: include: at least one processor (5) and memory (6); The memory (6) is used to store computer-executable instructions; The at least one processor (5) is used to execute the computer-executable instructions stored in the memory (6) to implement a plant growth prediction method based on a machine learning model as described in any one of claims 1 to 8.

Images