Alfalfa growth quality assessment method and system based on artificial intelligence

Abstract

The present invention discloses an alfalfa growth quality assessment method and system based on artificial intelligence, and the method includes data collection, data preprocessing, alfalfa growth environment assessment, alfalfa growth situation prediction and alfalfa growth quality assessment. The present invention relates to the field of alfalfa growth management technology, and specifically refers to an alfalfa growth quality assessment method and system based on artificial intelligence. The present invention adopts a two-way analysis combining alfalfa growth environment assessment and growth situation prediction, and combines the two to conduct a comprehensive assessment and judgment of alfalfa growth quality, thereby improving the depth and reference of alfalfa growth quality assessment; adopts a recursive long short-term memory network combined with data envelopment analysis to conduct alfalfa growth environment assessment, screens environmental data through envelope analysis, optimizes short-term data through recursive subnets, and optimizes long-term data analysis through long short-term memory subnets; adopts a gated recurrent unit model combined with masked region convolution to predict alfalfa growth situation, thereby improving the accuracy of prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of alfalfa growth management, and in particular to an alfalfa growth quality assessment method and system based on artificial intelligence. Background Art

[0002] The AI-based alfalfa growth quality assessment method is a method that uses machine learning and deep learning technologies to intelligently analyze and evaluate the alfalfa growth process. This method monitors and evaluates the growth status of alfalfa in real time by integrating multiple data sources, such as environmental conditions, plant growth parameters, and image data. Its main functions include improving the efficiency of alfalfa growth management, optimizing fertilization and irrigation strategies, and predicting possible growth problems, thereby achieving high quality and high efficiency of alfalfa production. This intelligent assessment method can automatically process large amounts of data, reduce manual intervention, and provide scientific and accurate growth quality assessment reports to help farmers and producers make better decisions.

[0003] However, in the existing alfalfa growth quality assessment methods, there is a technical problem that the traditional alfalfa growth quality assessment often focuses on the growth status of the plant itself, and adopts image-based analysis and growth assessment. Due to the complexity of the alfalfa growth environment, this method is difficult to combine with objective growth conditions for a deeper analysis and assessment, thereby reducing the overall assessment depth of alfalfa growth quality; in the existing alfalfa growth environment assessment methods, there is a technical problem that the alfalfa growth environment has multiple complex variables and factors, and data that is too complex and difficult to detect potential rules will interfere with the effect of the growth environment assessment. At the same time, the complex characteristics of the growth environment determine that the environmental analysis has both long-term subtle growth influences and immediate environmental influencing factors. Therefore, both real-time analysis and long-term analysis must be considered, and the alfalfa growth environment data must be effectively screened; in the existing alfalfa growth prediction methods, there is a technical problem that the traditional method is difficult to combine environmental data for effective spatiotemporal analysis based on a single image analysis. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an alfalfa growth quality assessment method and system based on artificial intelligence. In view of the fact that in the existing alfalfa growth quality assessment method, the traditional alfalfa growth quality assessment often focuses on the growth status of the plant itself, and adopts image-based analysis and growth assessment. Due to the complexity of the alfalfa growth environment, this method is difficult to combine objective growth conditions for a deeper analysis and assessment, thereby reducing the technical problem of the depth of the overall assessment of alfalfa growth quality. This scheme creatively adopts a two-way analysis combining alfalfa growth environment assessment and growth situation prediction, and combines the two to conduct a comprehensive assessment and judgment of alfalfa growth quality, thereby improving the depth and reference value of alfalfa growth quality assessment; in view of the fact that in the existing alfalfa growth environment assessment method, the alfalfa growth environment has multiple complex variables and factors, and data that is too complex and difficult to detect potential rules will interfere with the effect of the growth environment assessment. At the same time, the complex characteristics of the growth environment determine that the environmental analysis has both long-term subtle growth effects and immediate effects. environmental factors that affect the growth of alfalfa, so both real-time analysis and long-term analysis must be considered, and the technical problem of effectively screening the alfalfa growth environment data must be solved. This solution creatively uses a recursive long short-term memory network combined with data envelopment analysis to evaluate the growth environment of alfalfa, screens environmental data through improved envelope analysis, optimizes short-term data through recursive subnets, and optimizes the analysis of long-term impact data through long short-term memory subnets, thereby improving the task feasibility and result reliability of alfalfa growth environment evaluation; in view of the technical problem that traditional methods are difficult to combine single image analysis with environmental data for effective spatiotemporal analysis in the existing alfalfa growth prediction methods, this solution creatively uses a gated recurrent unit model combined with masked region convolution to predict alfalfa growth, and the masked region convolution realizes high-precision spatial information extraction, and the gated recurrent unit is used to simply and efficiently capture the time-dependent data in the growth prediction, thereby improving the ability to combine environmental analysis with alfalfa growth prediction, and comprehensively improving the accuracy of alfalfa growth prediction.

[0005] The technical solution adopted by the present invention is as follows: the alfalfa growth quality assessment method based on artificial intelligence provided by the present invention comprises the following steps:

[0006] Step S1: data collection;

[0007] Step S2: data preprocessing;

[0008] Step S3: alfalfa growth environment assessment;

[0009] Step S4: predicting alfalfa growth;

[0010] Step S5: Alfalfa growth quality assessment.

[0011] Furthermore, in step S1, the data collection is used to collect the original data required for the alfalfa growth quality assessment, specifically, to obtain the original data for the alfalfa growth quality assessment from the alfalfa natural growth environment through data collection;

[0012] The original data for evaluating alfalfa growth quality specifically include meteorological data, growth stage image data, insect and disease data, and soil nutrient component analysis data.

[0013] Furthermore, in step S2, the data preprocessing is used to perform basic optimization processing on the original data, specifically, data preprocessing and feature engineering design are performed on the original data of alfalfa growth quality assessment to obtain optimized data for alfalfa growth quality assessment.

[0014] Further, in step S3, the alfalfa growth environment assessment is used to pre-assess the growth environment of alfalfa, specifically, based on the alfalfa growth quality assessment optimization data, a recursive long short-term memory network combined with data envelopment analysis is used to conduct alfalfa growth environment assessment to obtain alfalfa growth environment assessment data;

[0015] The recursive long short-term memory network combined with data envelopment analysis specifically includes a data envelopment analysis operator, a recursive neural subnet and a long short-term memory subnet;

[0016] The data envelopment analysis operator is used to evaluate the efficiency of complex conditions in the alfalfa growth environment and pre-screen time series data to improve the accuracy and pertinence of the growth environment evaluation;

[0017] The recursive neural subnetwork is used to optimize the modeling performance of short-term environmental data in alfalfa growth environment data, and improve the analysis ability of environmental data that is volatile in a short period of time;

[0018] The long short-term memory subnet is used to optimize the modeling performance of long-term environmental data in alfalfa growth environment data, and to improve the analysis and processing capabilities of environmental factors that have a greater impact on the growth environment over a long period of time;

[0019] The step of using a recursive long short-term memory network combined with data envelopment analysis to evaluate the alfalfa growth environment and obtain alfalfa growth environment evaluation data includes:

[0020] Step S31: constructing a data envelopment analysis operator, specifically, taking the alfalfa growth environment characteristic data in the alfalfa growth quality evaluation optimization data as the input of the data envelopment analysis operator, and performing improved data envelopment analysis through efficiency value calculation and data screening, wherein the improved data envelopment analysis specifically optimizes the linear programming process by introducing dynamic weights and constraint conditions, and performs weighted calculation of efficiency scores through alfalfa growth hierarchical envelope analysis to obtain analysis results of the degree of influence of environmental variables on alfalfa growth, and based on the analysis results, screening alfalfa growth environment variables to obtain pre-screened alfalfa growth environment variable data;

[0021] Step S32: constructing a recursive neural subnetwork, specifically, inputting the pre-screened alfalfa growth environment variable data as time series data, constructing a recursive neural subnetwork, calculating the hidden state of the alfalfa growth environment at the corresponding time step, and obtaining the hidden state data output by the recursive neural subnetwork;

[0022] Step S33: constructing a long short-term memory subnet, specifically, constructing a standard long short-term memory neural network including an input gate, a forget gate and an output gate according to the hidden state data output by the recursive neural subnet, and performing long short-term memory subnet prediction by updating the hidden state of the long short-term memory subnet to obtain the hidden state of the long short-term memory subnet output, and generating a final prediction result by constructing a fully connected layer to obtain a prediction result of the alfalfa growth environment;

[0023] Step S34: Optimizing and training the environmental assessment model, specifically, adjusting the model parameters by using the data envelopment analysis operator, the recursive neural subnet and the long short-term memory subnet, optimizing and training the environmental assessment model, and obtaining the alfalfa growth environment assessment model Model DL ;

[0024] Step S35: Alfalfa growth environment assessment, specifically, using the alfalfa growth environment assessment model Model according to the alfalfa growth quality assessment optimization data DL , conduct alfalfa growth environment assessment and obtain alfalfa growth environment assessment data.

[0025] Further, in step S4, the alfalfa growth prediction is used to predict the alfalfa growth by combining historical data and the growth environment assessment data, specifically, based on the alfalfa growth environment assessment data and the alfalfa growth quality assessment optimization data, a gated recurrent unit model combined with masked region convolution is used to predict the alfalfa growth to obtain alfalfa growth prediction data;

[0026] The gated recurrent unit model combined with masked region convolution specifically includes a masked region convolution subnet and a gated recurrent unit subnet;

[0027] The masked region convolutional subnetwork is used to realize target detection and segmentation of images of alfalfa growth stages, and enhance the quantification and identification of the coverage area and leaf area index in the growth state of alfalfa;

[0028] The gated recurrent unit subnetwork is used to combine the image features extracted by the masked region convolution subnetwork to simplify the prediction of complex alfalfa growth data, optimize the time series data processing capability in the alfalfa growth environment assessment data, and realize the spatiotemporal information integration and refined prediction of image data and time series data;

[0029] The step of using the gated recurrent unit model combined with masked region convolution to predict alfalfa growth to obtain alfalfa growth prediction data includes:

[0030] Step S41: Growth data reconstruction, specifically using a pre-trained YOLOv8 model to pre-detect alfalfa areas, and perform data labeling and manual segmentation labeling to obtain alfalfa growth stage image data;

[0031] Step S42: constructing a masked region convolutional subnetwork, specifically constructing a masked region convolutional subnetwork including an input layer, a backbone network layer, a convolution, a lightweight region screening layer, and a region feature pooling layer, performing feature extraction and segmentation mask generation of an image of alfalfa growth stage, and obtaining masked feature data of alfalfa growth stage;

[0032] Step S43: constructing a gated recurrent unit subnet, specifically combining the alfalfa growth stage mask feature data and the alfalfa growth environment assessment data, taking the alfalfa growth environment assessment data as time series feature data, and taking the alfalfa growth stage mask feature data as image feature data, performing standard gated recurrent unit construction, obtaining a gated recurrent unit prediction output, and fusing the gated recurrent unit prediction output of the image feature data and the gated recurrent unit prediction output of the time series feature data by constructing a fully connected layer to obtain an alfalfa growth stage prediction output;

[0033] Step S44: training the alfalfa growth prediction model, specifically, reconstructing the growth data, constructing the masked region convolutional subnet and constructing the gated recurrent unit subnet, and using the joint loss function as the model loss function to train the alfalfa growth prediction model to obtain the alfalfa growth prediction model Model MG ;

[0034] Step S45: predicting the growth of alfalfa, specifically using the alfalfa growth prediction model Model according to the alfalfa growth environment assessment data and the alfalfa growth quality assessment optimization data MG , predict the growth of alfalfa and obtain the prediction data of alfalfa growth.

[0035] Furthermore, in step S5, the alfalfa growth quality assessment is used to make a comprehensive prediction of alfalfa growth quality by combining the growth environment assessment and the growth condition prediction. Specifically, the alfalfa growth environment assessment data and the alfalfa growth condition prediction data are combined to make a comprehensive quality prediction and assessment of alfalfa growth to obtain comprehensive analysis data of alfalfa growth quality.

[0036] The alfalfa growth quality assessment system based on artificial intelligence provided by the present invention comprises a data collection module, a data preprocessing module, an alfalfa growth environment assessment module, an alfalfa growth condition prediction module and an alfalfa growth quality assessment module;

[0037] The data collection module is used for data collection, and obtains raw data of alfalfa growth quality assessment through data collection, and sends the raw data of alfalfa growth quality assessment to the data preprocessing module;

[0038] The data preprocessing module is used for data preprocessing, and obtains alfalfa growth quality evaluation optimization data through data preprocessing, and sends the alfalfa growth quality evaluation optimization data to the alfalfa growth environment evaluation module;

[0039] The alfalfa growth environment assessment module is used for alfalfa growth environment assessment, obtains alfalfa growth environment assessment data through alfalfa growth environment assessment, and sends the alfalfa growth environment assessment data to the alfalfa growth condition prediction module and the alfalfa growth quality assessment module;

[0040] The alfalfa growth prediction module is used for alfalfa growth prediction, obtains alfalfa growth prediction data through alfalfa growth prediction, and sends the alfalfa growth prediction data to the alfalfa growth quality assessment module;

[0041] The alfalfa growth quality assessment module is used for alfalfa growth quality assessment, and through the alfalfa growth quality assessment, alfalfa growth quality comprehensive analysis data is obtained.

[0042] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0043] (1) In view of the existing alfalfa growth quality assessment methods, the traditional alfalfa growth quality assessment often focuses on the growth status of the plant itself, using image-based analysis and growth assessment. Due to the complexity of the alfalfa growth environment, this method is difficult to conduct a deeper analysis and assessment in combination with objective growth conditions, thereby reducing the technical problem of the depth of the overall assessment of alfalfa growth quality. This scheme creatively adopts a two-way analysis combining alfalfa growth environment assessment and growth condition prediction, and combines the two to conduct a comprehensive assessment and judgment of alfalfa growth quality, thereby improving the depth and reference value of alfalfa growth quality assessment;

[0044] (2) In the existing alfalfa growth environment assessment methods, there are multiple complex variables and factors in the alfalfa growth environment. Data that is too complex and difficult to detect underlying rules will interfere with the effect of the growth environment assessment. At the same time, the complex characteristics of the growth environment determine that environmental analysis has both long-term subtle growth influences and immediate environmental influence factors. Therefore, both real-time analysis and long-term analysis must be considered, and technical issues such as effective screening of alfalfa growth environment data must be addressed. This scheme creatively uses a recursive long short-term memory network combined with data envelopment analysis to conduct alfalfa growth environment assessment, screens environmental data through improved envelope analysis, optimizes short-term data through a recursive subnet, and optimizes the analysis of long-term impact data through a long short-term memory subnet, thereby improving the task feasibility and result reliability of alfalfa growth environment assessment.

[0045] (3) In view of the technical problem that the traditional methods of alfalfa growth prediction are difficult to combine with environmental data for effective spatiotemporal analysis based on single image analysis, this scheme creatively adopts a gated recurrent unit model combined with masked region convolution to predict alfalfa growth. The masked region convolution can achieve high-precision spatial information extraction, and the gated recurrent unit can simply and efficiently capture the time-dependent data in the growth prediction, thereby improving the ability to combine environmental analysis and alfalfa growth prediction, and comprehensively improving the accuracy of alfalfa growth prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of the process of the alfalfa growth quality assessment method based on artificial intelligence provided by the present invention;

[0047] Figure 2 A schematic diagram of an artificial intelligence-based alfalfa growth quality assessment system provided by the present invention;

[0048] Figure 3 This is a schematic diagram of the process of data preprocessing in step S2;

[0049] Figure 4 This is a schematic diagram of the process of evaluating the alfalfa growth environment in step S3;

[0050] Figure 5 This is a schematic diagram of the process of predicting alfalfa growth in step S4.

[0051] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0053] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0054] Example 1, see Figure 1 The invention provides an artificial intelligence-based alfalfa growth quality assessment method, which comprises the following steps:

[0055] Step S1: data collection;

[0056] Step S2: data preprocessing;

[0057] Step S3: alfalfa growth environment assessment;

[0058] Step S4: predicting alfalfa growth;

[0059] Step S5: Alfalfa growth quality assessment.

[0060] By performing the above operations, in view of the fact that in the existing alfalfa growth quality assessment methods, the traditional alfalfa growth quality assessment often focuses on the growth status of the plant itself, and adopts image-based analysis and growth assessment. Due to the complexity of the alfalfa growth environment, this method is difficult to combine objective growth conditions for further analysis and evaluation, thereby reducing the technical problem of the depth of the overall assessment of alfalfa growth quality. This solution creatively adopts a two-way analysis combining alfalfa growth environment assessment and growth condition prediction, and combines the two to conduct a comprehensive assessment and judgment of alfalfa growth quality, thereby improving the depth and reference value of alfalfa growth quality assessment.

[0061] Example 2, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S1, the data collection is used to collect the original data required for alfalfa growth quality assessment, specifically, to obtain the original data for alfalfa growth quality assessment from the natural growth environment of alfalfa through data collection;

[0062] The original data for evaluating alfalfa growth quality specifically include meteorological data, growth stage image data, insect and disease data, and soil nutrient component analysis data.

[0063] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the data preprocessing is used to perform basic optimization processing on the original data, specifically to perform data preprocessing and feature engineering design on the original data of alfalfa growth quality assessment to obtain optimized data for alfalfa growth quality assessment, and specifically includes the following steps:

[0064] Step S21: data cleaning, specifically detecting and processing missing values ​​and abnormal values ​​in the original data of alfalfa growth quality assessment to obtain alfalfa growth cleaning data;

[0065] Step S22: data standardization, specifically using Z-score standardization to perform standardization processing on the alfalfa growth cleaning data to obtain standardized alfalfa growth data;

[0066] Step S23: extracting growth features, specifically using a standard convolutional neural network combined with a target detection network to extract features from the image data in the alfalfa growth quality assessment raw data to obtain an image feature data set;

[0067] Step S24: environmental feature selection, specifically, extracting and selecting features from the standardized alfalfa growth data through manual labeling and manual feature extraction operations to obtain alfalfa growth environment feature data;

[0068] Step S25: data optimization processing, specifically, performing data preprocessing and feature engineering design on the original data of alfalfa growth quality assessment through the data cleaning, data standardization, growth feature extraction and environmental feature selection, so as to obtain optimized data for alfalfa growth quality assessment.

[0069] Example 4, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the alfalfa growth environment assessment is used to pre-assess the growth environment of alfalfa. Specifically, based on the alfalfa growth quality assessment optimization data, a recursive long short-term memory network combined with data envelopment analysis is used to conduct alfalfa growth environment assessment to obtain alfalfa growth environment assessment data;

[0070] The recursive long short-term memory network combined with data envelopment analysis specifically includes a data envelopment analysis operator, a recursive neural subnet and a long short-term memory subnet;

[0071] The data envelopment analysis operator is used to evaluate the efficiency of complex conditions in the alfalfa growth environment and pre-screen time series data to improve the accuracy and pertinence of the growth environment evaluation;

[0072] The recursive neural subnetwork is used to optimize the modeling performance of short-term environmental data in alfalfa growth environment data, and improve the analysis ability of environmental data that is volatile in a short period of time;

[0073] The long short-term memory subnet is used to optimize the modeling performance of long-term environmental data in alfalfa growth environment data, and to improve the analysis and processing capabilities of environmental factors that have a greater impact on the growth environment over a long period of time;

[0074] The step of using a recursive long short-term memory network combined with data envelopment analysis to evaluate the alfalfa growth environment and obtain alfalfa growth environment evaluation data includes:

[0075] Step S31: constructing a data envelopment analysis operator, specifically, taking the alfalfa growth environment characteristic data in the alfalfa growth quality evaluation optimization data as the input of the data envelopment analysis operator, and performing improved data envelopment analysis through efficiency value calculation and data screening, wherein the improved data envelopment analysis specifically optimizes the linear programming process by introducing dynamic weights and constraint conditions, and performs weighted calculation of efficiency scores through alfalfa growth hierarchical envelope analysis to obtain analysis results of the degree of influence of environmental variables on alfalfa growth, and based on the analysis results, screening alfalfa growth environment variables to obtain pre-screened alfalfa growth environment variable data;

[0076] The calculation formula for screening the alfalfa growth environment variables is:

[0077] ;

[0078] Where S is the pre-screening alfalfa growth environment variable data, It is alfalfa growth environment variables, the alfalfa growth environment variables are specifically obtained through data screening, including input indicators and output indicators, is the total efficiency score, which is calculated by improved data envelopment analysis. is the screening threshold, specifically the screening threshold of the total efficiency score, which is 0.75. When the total efficiency score corresponding to the alfalfa growth environment variable is greater than or equal to the screening threshold, the alfalfa growth environment variable is retained in the pre-screening alfalfa growth environment variable data. When the total efficiency score corresponding to an alfalfa growth environment variable is less than the screening threshold, the alfalfa growth environment variable is eliminated;

[0079] The data screening specifically includes input index screening and output index screening, the input index screening specifically includes temperature, humidity, precipitation and soil fertility data, and the output index screening specifically includes growth rate prediction value and yield prediction value;

[0080] The efficiency value calculation is specifically carried out by introducing dynamic weights and linear programming of constraint optimization to calculate the relative efficiency of decision-making units, and performing weighted calculation of efficiency scores through hierarchical envelope analysis of alfalfa growth;

[0081] The decision-making unit specifically includes a growth rate prediction decision-making unit and a yield prediction decision-making unit;

[0082] The alfalfa growth stages specifically include the germination stage, the maturity stage and the harvest stage;

[0083] The step of performing improved data envelopment analysis comprises:

[0084] Step S311: Dynamically adjust the linear programming optimization, specifically by introducing the time factor to reconstruct the objective function of the linear programming. The calculation formula is:

[0085] ;

[0086] In the formula, is the objective function of reconstructed linear programming, Max is the maximum value function, is the efficiency score of the jth decision unit at time step t, j is the decision unit index, s is the total number of output variables, the specific value is 2, r is the output variable index, the specific value range is {1,2}, where the value is 1, indicating the growth rate output, and the value is 2, indicating the output output. The output variables specifically include the growth rate prediction value and the yield prediction value, a t is the time weight factor, which is used to dynamically adjust the weight value of the environment variable. is the time-varying output weight of the rth output variable at time step t, is the specific output value of the rth output variable of the jth decision unit;

[0087] Step S312: construct dynamic adjustment constraints, specifically by introducing a time factor to adjust the standardization constraints of the input data. The calculation formula is:

[0088] ;

[0089] In the formula, The overall constraint is dynamically adjusted. M is the total number of input variables, with a specific value of 4. i is the input variable index, with a specific value range of {1, 2, 3, 4}. When the value is 1, it indicates temperature data; when the value is 2, it indicates humidity data; when the value is 3, it indicates precipitation data; and when the value is 4, it indicates soil fertility data. is the time-varying input weight of the ith input variable at time step t, a t is the time weighting factor, is the specific input value of the i-th input variable of the j-th decision unit;

[0090] Step S313: construct upper and lower boundary constraints to optimize the adaptability of data envelopment analysis to different environmental variables and reconstruct the upper and lower boundary constraint ranges of input variables. The calculation formula is:

[0091] ;

[0092] Where, L i is the lower bound of the ith input variable, U i is the upper bound of the ith input variable, is the time-varying input weight of the ith input variable at time step t, is the specific input value of the i-th input variable of the j-th decision unit;

[0093] Step S314: hierarchical data envelopment analysis, specifically, constructing hierarchical sub-decision making units according to the stages of alfalfa growth process, and calculating the final total efficiency score by weighted summation, the calculation formula is:

[0094] ;

[0095] In the formula, is the total efficiency score, P is the total number of hierarchical stages, the specific value is 3, p is the hierarchical stage index, which is used to indicate the specific number of hierarchical sub-decision-making units, and the value range is {1,2,3}. When the value is 1, it indicates the germination stage, when the value is 2, it indicates the growth stage, and when the value is 3, it indicates the harvest stage. p is the stage weight of the pth hierarchical stage, is the efficiency score of the jth decision-making unit at the pth hierarchical stage;

[0096] Step S32: constructing a recursive neural subnetwork, specifically, inputting the pre-screened alfalfa growth environment variable data as time series data, constructing a recursive neural subnetwork, calculating the hidden state of the alfalfa growth environment at the corresponding time step, and obtaining the hidden state data output by the recursive neural subnetwork;

[0097] The recursive neural subnet is constructed by using a standard recursive neural network, and the standard recursive neural network includes an input layer, a hidden layer and an output layer.

[0098] Table 1 is a schematic table of parameters of the standard recursive neural network. As shown in the table, the schematic parameters of the standard recursive neural network specifically include input dimension, hidden layer dimension, output dimension, time step number, batch size, learning rate, optimization algorithm, activation function and loss function;

[0099] The specific value of the input dimension is 10; the specific value of the hidden layer dimension is 50; the specific value of the output dimension is 50; the specific value of the time step is 100; the specific value of the batch size is 32; the specific value of the learning rate is 0.001; the optimization algorithm specifically adopts the Adam algorithm; the activation function specifically adopts the hyperbolic tangent activation function; the loss function specifically adopts the cross entropy loss function;

[0100] Table 1 Parameters of the standard recurrent neural network

[0101]

[0102] Step S33: constructing a long short-term memory subnet, specifically, constructing a standard long short-term memory neural network including an input gate, a forget gate and an output gate according to the hidden state data output by the recursive neural subnet, and performing long short-term memory subnet prediction by updating the hidden state of the long short-term memory subnet to obtain the hidden state of the long short-term memory subnet output, and generating a final prediction result by constructing a fully connected layer to obtain a prediction result of the alfalfa growth environment;

[0103] Step S34: Optimizing and training the environmental assessment model, specifically, adjusting the model parameters by using the data envelopment analysis operator, the recursive neural subnet and the long short-term memory subnet, optimizing and training the environmental assessment model, and obtaining the alfalfa growth environment assessment model Model DL ;

[0104] Step S35: Alfalfa growth environment assessment, specifically, using the alfalfa growth environment assessment model Model according to the alfalfa growth quality assessment optimization data DL , conduct alfalfa growth environment assessment and obtain alfalfa growth environment assessment data.

[0105] By performing the above operations, in the existing alfalfa growth environment assessment methods, there are multiple complex variables and factors in the alfalfa growth environment. Data that is too complex and difficult to detect potential rules will interfere with the effect of the growth environment assessment. At the same time, the complex characteristics of the growth environment determine that the environmental analysis has both long-term subtle growth influences and immediate environmental influence factors. Therefore, both real-time analysis and long-term analysis must be considered, and the technical problem of effectively screening alfalfa growth environment data must be solved. This solution creatively uses a recursive long short-term memory network combined with data envelopment analysis to evaluate the alfalfa growth environment, screens environmental data through improved envelope analysis, optimizes short-term data through recursive subnets, and optimizes the analysis of long-term impact data through long short-term memory subnets, thereby improving the task feasibility and result reliability of alfalfa growth environment assessment.

[0106] Example 5, see Figure 1 , Figure 2 and Figure 5 This embodiment is based on the above embodiment. In step S4, the alfalfa growth prediction is used to predict the alfalfa growth by combining historical data and the growth environment assessment data. Specifically, the alfalfa growth prediction is performed based on the alfalfa growth environment assessment data and the alfalfa growth quality assessment optimization data, using a gated recurrent unit model combined with masked region convolution to obtain alfalfa growth prediction data;

[0107] The gated recurrent unit model combined with masked region convolution specifically includes a masked region convolution subnet and a gated recurrent unit subnet;

[0108] The masked region convolutional subnetwork is used to realize target detection and segmentation of images of alfalfa growth stages, and enhance the quantification and identification of the coverage area and leaf area index in the growth state of alfalfa;

[0109] The gated recurrent unit subnetwork is used to combine the image features extracted by the masked region convolution subnetwork to simplify the prediction of complex alfalfa growth data, optimize the time series data processing capability in the alfalfa growth environment assessment data, and realize the spatiotemporal information integration and refined prediction of image data and time series data;

[0110] The step of using the gated recurrent unit model combined with masked region convolution to predict alfalfa growth to obtain alfalfa growth prediction data includes:

[0111] Step S41: Growth data reconstruction, specifically using a pre-trained YOLOv8 model to pre-detect alfalfa areas, and perform data labeling and manual segmentation labeling to obtain alfalfa growth stage image data;

[0112] Step S42: constructing a masked region convolutional subnetwork, specifically constructing a masked region convolutional subnetwork including an input layer, a backbone network layer, a convolution, a lightweight region screening layer, and a region feature pooling layer, performing feature extraction and segmentation mask generation of an image of alfalfa growth stage, and obtaining masked feature data of alfalfa growth stage;

[0113] The lightweight region screening layer specifically introduces a dynamic mask generation mechanism to fuse image inputs at different time points with the mask at the current time, and dynamically adjusts the mask generation process by introducing a time weight coefficient;

[0114] The regional feature pooling layer specifically extracts environmental information regional features from the alfalfa growth environment assessment data by introducing a context perception module, and enhances the accuracy of mask features by a maximum pooling operation;

[0115] Table 2 is a parameter schematic table of the masked area convolutional subnet, as shown in Table 2, the schematic parameters of the masked area convolutional subnet specifically include input dimension, backbone network, convolution kernel size, activation function, mask generation time weight coefficient, pooling method, context perception module and output dimension;

[0116] The input dimension is specifically set to 256×256×3, which is used to represent the size of the image data of the alfalfa growth stage as the input image; the backbone network is specifically a pre-trained ResNet50 network; the convolution kernel size is specifically set to 3×3; the activation function is specifically set to ReLU; the mask generation time weight coefficient is specifically set to 0.5; the pooling method is specifically a maximum pooling operation; the context perception module is specifically a global context perception module; the output dimension is specifically set to 256×256×1, which is used to represent the final mask feature size of the output mask feature data of the alfalfa growth stage;

[0117] Table 2 Parameters of the masked region convolutional subnet

[0118]

[0119] Step S43: constructing a gated recurrent unit subnet, specifically combining the alfalfa growth stage mask feature data and the alfalfa growth environment assessment data, taking the alfalfa growth environment assessment data as time series feature data, and taking the alfalfa growth stage mask feature data as image feature data, performing standard gated recurrent unit construction, obtaining a gated recurrent unit prediction output, and fusing the gated recurrent unit prediction output of the image feature data and the gated recurrent unit prediction output of the time series feature data by constructing a fully connected layer to obtain an alfalfa growth stage prediction output;

[0120] Step S44: training the alfalfa growth prediction model, specifically, reconstructing the growth data, constructing the masked region convolutional subnet and constructing the gated recurrent unit subnet, and using the joint loss function as the model loss function to train the alfalfa growth prediction model to obtain the alfalfa growth prediction model Model MG ;

[0121] The calculation formula of the joint loss function is:

[0122] ;

[0123] Where, L Total is the joint loss function, is the loss weight of the masked region convolutional subnet, L MRC is the masked region convolutional subnet loss function, is the gated recurrent unit subnetwork loss weight, L GRU is the gated recurrent unit subnetwork loss function, specifically the mean square error loss function, L Cls is the classification loss function, L Bbox is the bounding box loss function, L Mask is the enhanced alfalfa region mask loss function, T is the total time step of the alfalfa region mask loss dynamicization, is the time step index of the dynamic mask loss of the alfalfa region, is the dynamic time coefficient of the mask loss in the alfalfa region, is the time step The mask loss function when ;

[0124] Step S45: predicting the growth of alfalfa, specifically using the alfalfa growth prediction model Model according to the alfalfa growth environment assessment data and the alfalfa growth quality assessment optimization data MG , predict the growth of alfalfa and obtain the prediction data of alfalfa growth.

[0125] By performing the above operations, in order to address the technical problem in the existing alfalfa growth prediction methods that traditional methods are difficult to combine single image analysis with environmental data for effective spatiotemporal analysis, this scheme creatively adopts a gated recurrent unit model combined with masked region convolution to predict alfalfa growth. The masked region convolution achieves high-precision spatial information extraction, and the gated recurrent unit is used to simply and efficiently capture the time-dependent data in the growth prediction, thereby improving the ability to combine environmental analysis and alfalfa growth prediction, and comprehensively improving the accuracy of alfalfa growth prediction.

[0126] Example 6, see Figure 1 and Figure 2This embodiment is based on the above embodiment. In step S5, the alfalfa growth quality assessment is used to comprehensively predict the growth quality of alfalfa by combining the growth environment assessment and the growth situation prediction. Specifically, the alfalfa growth comprehensive quality prediction and assessment are performed by combining the alfalfa growth environment assessment data and the alfalfa growth situation prediction data to obtain alfalfa growth quality comprehensive analysis data.

[0127] The calculation formula for the comprehensive quality prediction and evaluation of alfalfa growth is:

[0128] ;

[0129] Where, Q is the comprehensive analysis data of alfalfa growth quality, which is used to make a comprehensive quantitative expression of alfalfa growth quality, and W DL is the growth environment assessment weight, DL is the alfalfa growth environment assessment data, W MG is the growth prediction weight, and MG is the alfalfa growth prediction data.

[0130] Embodiment 7, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The artificial intelligence-based alfalfa growth quality assessment system provided by the present invention includes a data collection module, a data preprocessing module, an alfalfa growth environment assessment module, an alfalfa growth condition prediction module and an alfalfa growth quality assessment module;

[0131] The data collection module is used for data collection, and obtains raw data of alfalfa growth quality assessment through data collection, and sends the raw data of alfalfa growth quality assessment to the data preprocessing module;

[0132] The data preprocessing module is used for data preprocessing, and obtains alfalfa growth quality evaluation optimization data through data preprocessing, and sends the alfalfa growth quality evaluation optimization data to the alfalfa growth environment evaluation module;

[0133] The alfalfa growth environment assessment module is used for alfalfa growth environment assessment, obtains alfalfa growth environment assessment data through alfalfa growth environment assessment, and sends the alfalfa growth environment assessment data to the alfalfa growth condition prediction module and the alfalfa growth quality assessment module;

[0134] The alfalfa growth prediction module is used for alfalfa growth prediction, obtains alfalfa growth prediction data through alfalfa growth prediction, and sends the alfalfa growth prediction data to the alfalfa growth quality assessment module;

[0135] The alfalfa growth quality assessment module is used for alfalfa growth quality assessment, and through the alfalfa growth quality assessment, alfalfa growth quality comprehensive analysis data is obtained.

[0136] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0137] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0138] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. An artificial intelligence-based alfalfa growth quality assessment method, characterized in that: The method comprises the following steps: Step S1: data collection, obtaining raw data for alfalfa growth quality assessment; Step S2: data preprocessing to obtain alfalfa growth quality assessment optimization data; Step S3: Alfalfa growth environment assessment, used for pre-assessing the growth environment of alfalfa, specifically, based on the alfalfa growth quality assessment optimization data, adopting a recursive long short-term memory network combined with data envelopment analysis to conduct alfalfa growth environment assessment, and obtain alfalfa growth environment assessment data, specifically including the following steps: step S31: constructing a data envelopment analysis operator; step S32: constructing a recursive neural subnet; step S33: constructing a long short-term memory subnet; step S34: environmental assessment model optimization and training; step S35: alfalfa growth environment assessment; The data envelopment analysis operator is used to evaluate the efficiency of complex conditions in the alfalfa growth environment and pre-screen time series data to improve the accuracy and pertinence of the growth environment evaluation; The recursive neural subnetwork is used to optimize the modeling performance of short-term environmental data in alfalfa growth environment data, and improve the analysis ability of environmental data that is volatile in a short period of time; The long short-term memory subnet is used to optimize the modeling performance of long-term environmental data in alfalfa growth environment data, and to improve the analysis and processing capabilities of environmental factors that have a greater impact on the growth environment over a long period of time; Step S4: prediction of alfalfa growth, used to predict alfalfa growth by combining historical data and growth environment assessment data, specifically, based on the alfalfa growth environment assessment data and the alfalfa growth quality assessment optimization data, a gated recurrent unit model combined with masked region convolution is used to predict alfalfa growth, to obtain alfalfa growth prediction data, specifically including the following steps: step S41: growth data reconstruction; step S42: constructing a masked region convolution subnet; step S43: constructing a gated recurrent unit subnet; step S44: alfalfa growth prediction model training; step S45: alfalfa growth prediction; The masked region convolutional subnetwork is used to realize target detection and segmentation of images of alfalfa growth stages, and enhance the quantification and identification of the coverage area and leaf area index in the growth state of alfalfa; The gated recurrent unit subnet is used to combine the image features extracted by the masked region convolution subnet to simplify the prediction of complex alfalfa growth data, optimize the time series data processing capability in the alfalfa growth environment assessment data, and realize the spatiotemporal information integration and refined prediction of image data and time series data. Step S5: Alfalfa growth quality assessment is used to comprehensively predict the growth quality of alfalfa by combining the growth environment assessment and the growth situation prediction. Specifically, the comprehensive quality prediction and assessment of alfalfa growth are performed by combining the alfalfa growth environment assessment data and the alfalfa growth situation prediction data to obtain comprehensive analysis data of alfalfa growth quality.

2. The method for evaluating alfalfa growth quality based on artificial intelligence according to claim 1, characterized in that: In step S3, the step of using a recursive long short-term memory network combined with data envelopment analysis to evaluate the alfalfa growth environment and obtain alfalfa growth environment evaluation data includes: Step S31: constructing a data envelopment analysis operator, specifically, taking the alfalfa growth environment characteristic data in the alfalfa growth quality evaluation optimization data as the input of the data envelopment analysis operator, and performing improved data envelopment analysis through efficiency value calculation and data screening, wherein the improved data envelopment analysis specifically optimizes the linear programming process by introducing dynamic weights and constraint conditions, and performs weighted calculation of efficiency scores through alfalfa growth hierarchical envelope analysis to obtain analysis results of the degree of influence of environmental variables on alfalfa growth, and based on the analysis results, screening alfalfa growth environment variables to obtain pre-screened alfalfa growth environment variable data; Step S32: constructing a recursive neural subnetwork, specifically, inputting the pre-screened alfalfa growth environment variable data as time series data, constructing a recursive neural subnetwork, calculating the hidden state of the alfalfa growth environment at the corresponding time step, and obtaining the hidden state data output by the recursive neural subnetwork; Step S33: constructing a long short-term memory subnet, specifically, constructing a standard long short-term memory neural network including an input gate, a forget gate and an output gate according to the hidden state data output by the recursive neural subnet, and performing long short-term memory subnet prediction by updating the hidden state of the long short-term memory subnet to obtain the hidden state of the long short-term memory subnet output, and generating a final prediction result by constructing a fully connected layer to obtain a prediction result of the alfalfa growth environment; Step S34: Optimizing and training the environmental assessment model, specifically, adjusting the model parameters by using the data envelopment analysis operator, the recursive neural subnet and the long short-term memory subnet, optimizing and training the environmental assessment model, and obtaining the alfalfa growth environment assessment model Model DL ; Step S35: Alfalfa growth environment assessment, specifically, using the alfalfa growth environment assessment model Model according to the alfalfa growth quality assessment optimization data DL , conduct alfalfa growth environment assessment and obtain alfalfa growth environment assessment data.

3. The method for evaluating alfalfa growth quality based on artificial intelligence according to claim 2, characterized in that: In step S41, the growth data is reconstructed, specifically using a pre-trained YOLOv8 model to pre-detect alfalfa areas, and performing data labeling and manual segmentation labeling to obtain alfalfa growth stage image data; In step S42, the masked region convolutional subnetwork is constructed, specifically, a masked region convolutional subnetwork including an input layer, a backbone network layer, a convolution, a lightweight region screening layer, and a region feature pooling layer is constructed to extract features of the alfalfa growth stage image and generate a segmentation mask to obtain masked feature data of the alfalfa growth stage; In step S43, the construction of the gated recurrent unit subnetwork is specifically to combine the alfalfa growth stage mask feature data and the alfalfa growth environment assessment data, take the alfalfa growth environment assessment data as time series feature data, and take the alfalfa growth stage mask feature data as image feature data, perform standard gated recurrent unit construction, obtain a gated recurrent unit prediction output, and fuse the gated recurrent unit prediction output of the image feature data and the gated recurrent unit prediction output of the time series feature data by constructing a fully connected layer to obtain an alfalfa growth stage prediction output; In step S44, the alfalfa growth prediction model training is specifically carried out by reconstructing the growth data, constructing the mask region convolutional subnet and constructing the gated recurrent unit subnet, and using the mean square error loss function as the model loss function to train the alfalfa growth prediction model to obtain the alfalfa growth prediction model Model MG ; In step S45, the alfalfa growth condition prediction is specifically based on the alfalfa growth environment assessment data and the alfalfa growth quality assessment optimization data, using the alfalfa growth condition prediction model Model MG , predict the growth of alfalfa and obtain the prediction data of alfalfa growth.

4. An artificial intelligence-based alfalfa growth quality assessment system, used to implement the artificial intelligence-based alfalfa growth quality assessment method as described in any one of claims 1 to 3, characterized in that: It includes a data collection module, a data preprocessing module, an alfalfa growth environment assessment module, an alfalfa growth condition prediction module and an alfalfa growth quality assessment module; The data collection module is used for data collection, and obtains raw data of alfalfa growth quality assessment through data collection, and sends the raw data of alfalfa growth quality assessment to the data preprocessing module; The data preprocessing module is used for data preprocessing, and obtains alfalfa growth quality evaluation optimization data through data preprocessing, and sends the alfalfa growth quality evaluation optimization data to the alfalfa growth environment evaluation module; The alfalfa growth environment assessment module is used for alfalfa growth environment assessment, obtains alfalfa growth environment assessment data through alfalfa growth environment assessment, and sends the alfalfa growth environment assessment data to the alfalfa growth condition prediction module and the alfalfa growth quality assessment module; The alfalfa growth prediction module is used for alfalfa growth prediction, obtains alfalfa growth prediction data through alfalfa growth prediction, and sends the alfalfa growth prediction data to the alfalfa growth quality assessment module; The alfalfa growth quality assessment module is used for alfalfa growth quality assessment, and through the alfalfa growth quality assessment, alfalfa growth quality comprehensive analysis data is obtained.

Images