Sweet potato weevil influence distribution analysis platform based on simulation analysis

Abstract

The invention discloses a simulation analysis-based small weevil influence distribution analysis platform, which comprises the following steps of: setting multiple time nodes and sampling points in a historical monitoring time period, and analyzing first evaluation data of small weevil; combining the sweet potato image data with a comparison database, analyzing insect pest states through SSIM similarity, and obtaining second evaluation data; integrating the two types of data to generate a monitoring growth sequence based on a pest proportion, performing feature learning on the growth sequence, and training a GAN generation model; and collecting a real-time growth sequence, inputting the real-time growth sequence into the generative model to obtain a simulation sequence, combining the real-time growth sequence and the simulation sequence, predicting the growth trend of the sweet potato small weevil by using an LSTM network, and dynamically formulating a targeted prevention and control scheme and a monitoring scheme. The method can dynamically predict and analyze the growth state and quantity change of the sweet potato small weevil, and effectively improves the sweet potato pest monitoring and control efficiency.

Description

Technical Field

[0001] The present invention relates to the field of agricultural intelligent analysis, and more specifically, to a sweet potato weevil impact distribution analysis platform based on simulation analysis. Background Art

[0002] The sweet potato weevil is a major pest in sweet potato cultivation, severely impacting sweet potato yield and quality. In large-scale or multi-crop sweet potato cultivation environments, traditional technologies often struggle to accurately and efficiently analyze and predict the impact of the weevil, often relying on manual experience. Furthermore, in actual agricultural production, due to factors such as high monitoring costs, long monitoring cycles, and complex environmental conditions, it is often difficult to obtain sufficient pest monitoring data with sufficient quantity and coverage. This limits accurate assessments of pest growth patterns and the extent of damage, and makes it difficult to conduct long-term and efficient pest prediction and analysis, as well as simulate the growth trends of the weevil in complex environments. Furthermore, traditional pest control techniques for the weevil are often based on repetitive, single-type control solutions, which can easily lead to drug resistance or a certain degree of adaptability in field sweet potato pests. These technologies lack the ability to conduct dynamic pest analysis and dynamic solution development, and rely heavily on manual experience-based analysis and control, making them impractical and difficult to implement, further hindering the high-level development of sweet potato cultivation. Summary of the Invention

[0003] The present invention overcomes the defects of the prior art and proposes a sweet potato weevil impact distribution analysis platform based on simulation analysis.

[0004] A first aspect of the present invention provides a method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis, comprising:

[0005] S101: In a historical monitoring period, multiple monitoring time nodes are set, multiple sampling points are set based on a target sweet potato planting area, and first evaluation data based on the sweet potato weevil is obtained at the sampling points;

[0006] S102: acquiring sweet potato image data based on the sampling points, performing SSIM image similarity analysis based on a comparison database, performing pest assessment based on the similarity, and obtaining second assessment data based on the sweet potato weevil;

[0007] S103: Analyzing the ratio of adults to larvae using the first and second evaluation data, and sorting the ratio values ​​based on multiple nodes to obtain a monitoring growth sequence, constructing a generative model based on a GAN, using the monitoring growth sequence as training data, and performing cyclic adversarial training on the generative model until the model reaches a Nash equilibrium;

[0008] S104: Based on the real-time monitoring cycle, the ratio of adults to larvae in the target sweet potato planting area is analyzed, and a real-time growth sequence is generated. The real-time growth sequence is introduced into the generative model for sequence learning and generation of a simulation sequence;

[0009] S105: Sequence combination is performed based on the real-time growth sequence and the simulation sequence, and sequence prediction is performed on the combined sequence through the LSTM network to obtain a growth prediction sequence based on the sweet potato weevil, and data analysis is performed on the growth prediction sequence to set a corresponding prevention and control plan.

[0010] In this solution, the S101 is specifically:

[0011] Based on a historical monitoring period, multiple monitoring time nodes are set to ensure that the time intervals between two adjacent nodes are the same;

[0012] Based on the size of the target sweet potato area, multiple sampling points are set, and multiple sweet potato plants are sampled and pests are assessed at the sampling points to obtain first assessment data;

[0013] Pest assessment includes recording information on the number of adults, larvae, developmental stages, and damage symptoms of the sweet potato weevil.

[0014] In this solution, the S102 is specifically:

[0015] The image of the sampling point is acquired by the monitoring unit to obtain the sweet potato image data;

[0016] The sweet potato image data is preprocessed by image smoothing and noise reduction, and the sweet potato parts are segmented by edge detection operators to obtain multiple sweet potato part images;

[0017] Obtain comparative images of various parts of sweet potatoes under different growth conditions of sweet potato weevils from the comparison database. Unify the image size and color space of the sweet potato partial images and the corresponding comparison images. Set a 10x10 pixel window and calculate the similarity of each window in the form of SSIM by sliding the window. Finally, complete the calculation of the entire image area and obtain the SSIM mean.

[0018] The similarity between some sweet potato images and comparison images was analyzed by SSIM mean, and the pest status of sweet potato elephants was evaluated based on the similarity analysis of multiple comparison images;

[0019] SSIM similarity analysis and comparison were performed on multiple partial images of sweet potatoes, and a comprehensive pest assessment was conducted on the entire planting area based on the sampling points to generate the second assessment data.

[0020] In this solution, the S103 is specifically:

[0021] Performing adult-larva ratio analysis and numerical extraction based on the first evaluation data and the second evaluation data to obtain a numerical ratio for each node;

[0022] Based on the first evaluation data, serializing the proportional values ​​of N consecutive nodes to obtain a sequence segment, and generating multiple sequence segments based on multiple nodes;

[0023] generating a corresponding plurality of sequence segments based on the second evaluation data;

[0024] The sequence segments corresponding to the first and second evaluation data are used as monitoring growth sequences, and each sequence segment is vectorized to generate multiple feature vectors;

[0025] Build a GAN-based generative model, which includes a generator, a discriminator, a loss function, and an optimizer;

[0026] Import multiple feature vectors as real data into the generator for self-learning, and generate simulated features through the generator;

[0027] In the discriminator, the loss function is set based on the mean square error between the simulated features and the real data, and the accuracy of the simulated features is determined;

[0028] The discriminator and the generator are trained in an adversarial manner until they reach Nash equilibrium. The convergence value of the loss function and the optimizer status at this time are recorded to complete the model training.

[0029] In this solution, the S104 is specifically:

[0030] During a real-time monitoring cycle, we conducted an assessment of sweet potato weevil pests based on multiple sampling points, analyzed the ratio of adults to larvae at multiple nodes, and serialized the ratio values ​​to obtain a real-time growth sequence.

[0031] The real-time growth sequence is imported into the GAN model for sequence feature learning, and simulated data is generated based on the generator cycle. The discriminator extracts the simulated data with a probability greater than the preset discrimination probability to obtain the optimal simulated data. The simulation generation is repeated until the preset amount of optimal simulated data is obtained.

[0032] The optimized simulation data is subjected to data analysis to obtain a simulation sequence.

[0033] In this solution, the S105 is specifically:

[0034] Sequentially combine the real-time growth sequence and the simulation sequence to obtain a combined sequence;

[0035] The LSTM network is used to analyze and predict the sequence characteristics of the combined sequence, set the preset time nodes, and obtain the growth prediction sequence;

[0036] Data analysis of growth prediction sequences was performed to evaluate the growth trends and stages of the sweet potato weevil, predict pest distribution and set pest early warning information, and dynamically set prevention and control plans and monitoring plans based on the ratio of adults to larvae.

[0037] A second aspect of the present invention further provides a sweet potato weevil impact distribution analysis platform based on simulation analysis, the platform comprising: a memory and a processor, wherein the memory includes a sweet potato weevil impact distribution analysis program based on simulation analysis, and when the sweet potato weevil impact distribution analysis program based on simulation analysis is executed by the processor, the following steps are implemented:

[0038] S101: In a historical monitoring period, multiple monitoring time nodes are set, multiple sampling points are set based on a target sweet potato planting area, and first evaluation data based on the sweet potato weevil is obtained at the sampling points;

[0039] S102: acquiring sweet potato image data based on the sampling points, performing SSIM image similarity analysis based on a comparison database, performing pest assessment based on the similarity, and obtaining second assessment data based on the sweet potato weevil;

[0040] S103: Analyzing the ratio of adults to larvae using the first and second evaluation data, and sorting the ratio values ​​based on multiple nodes to obtain a monitoring growth sequence, constructing a generative model based on a GAN, using the monitoring growth sequence as training data, and performing cyclic adversarial training on the generative model until the model reaches a Nash equilibrium;

[0041] S104: Based on the real-time monitoring cycle, the ratio of adults to larvae in the target sweet potato planting area is analyzed, and a real-time growth sequence is generated. The real-time growth sequence is introduced into the generative model for sequence learning and generation of a simulation sequence;

[0042] S105: Sequence combination is performed based on the real-time growth sequence and the simulation sequence, and sequence prediction is performed on the combined sequence through the LSTM network to obtain a growth prediction sequence based on the sweet potato weevil, and data analysis is performed on the growth prediction sequence to set a corresponding prevention and control plan.

[0043] The third aspect of the present invention also provides a computer-readable storage medium, which includes a sweet potato weevil impact distribution analysis program based on simulation analysis. When the sweet potato weevil impact distribution analysis program based on simulation analysis is executed by a processor, it implements the steps of the sweet potato weevil impact distribution analysis method based on simulation analysis as described in any one of the above items.

[0044] The present invention discloses a sweet potato weevil impact distribution analysis platform based on simulation analysis, comprising: setting multiple time nodes and sampling points within a historical monitoring period and analyzing first assessment data of the sweet potato weevil; combining sweet potato image data with a comparison database, analyzing pest status through SSIM similarity, and obtaining second assessment data; integrating the two types of data to generate a monitoring growth sequence based on pest proportion, performing feature learning on the growth sequence and training a GAN generation model; collecting real-time growth sequences, inputting them into a generation model to obtain simulation sequences, combining the real-time growth sequences with the simulation sequences, using an LSTM network to predict the growth trend of the sweet potato weevil, and dynamically formulating targeted prevention and control plans and monitoring plans. The present invention can dynamically predict and analyze the growth status and number changes of the sweet potato weevil, effectively improving the efficiency of sweet potato pest monitoring and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flow chart of a method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis is shown in the present invention;

[0046] Figure 2 A block diagram of a sweet potato weevil impact distribution analysis platform based on simulation analysis of the present invention is shown. DETAILED DESCRIPTION

[0047] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0049] Figure 1 The present invention shows a flow chart of a method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis.

[0050] like Figure 1 As shown, the first aspect of the present invention provides a method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis, comprising:

[0051] S101: In a historical monitoring period, multiple monitoring time nodes are set, multiple sampling points are set based on a target sweet potato planting area, and first evaluation data based on the sweet potato weevil is obtained at the sampling points;

[0052] S102: acquiring sweet potato image data based on the sampling points, performing SSIM image similarity analysis based on a comparison database, performing pest assessment based on the similarity, and obtaining second assessment data based on the sweet potato weevil;

[0053] S103: Analyzing the ratio of adults to larvae using the first and second evaluation data, and sorting the ratio values ​​based on multiple nodes to obtain a monitoring growth sequence, constructing a generative model based on a GAN, using the monitoring growth sequence as training data, and performing cyclic adversarial training on the generative model until the model reaches a Nash equilibrium;

[0054] S104: Based on the real-time monitoring cycle, the ratio of adults to larvae in the target sweet potato planting area is analyzed, and a real-time growth sequence is generated. The real-time growth sequence is introduced into the generative model for sequence learning and generation of a simulation sequence;

[0055] S105: Sequence combination is performed based on the real-time growth sequence and the simulation sequence, and sequence prediction is performed on the combined sequence through the LSTM network to obtain a growth prediction sequence based on the sweet potato weevil, and data analysis is performed on the growth prediction sequence to set a corresponding prevention and control plan.

[0056] It should be noted that, in the present invention, a node refers to a monitoring time node.

[0057] According to an embodiment of the present invention, the S101 is specifically as follows:

[0058] Based on a historical monitoring period, multiple monitoring time nodes are set to ensure that the time intervals between two adjacent nodes are the same;

[0059] Based on the size of the target sweet potato area, multiple sampling points are set, and multiple sweet potato plants are sampled and pests are assessed at the sampling points to obtain first assessment data;

[0060] Pest assessment includes recording information on the number of adults, larvae, developmental stages, and damage symptoms of the sweet potato weevil.

[0061] It should be noted that the historical monitoring period is a longer monitoring cycle, which is used to analyze the growth patterns of sweet potato weevils in the planting area over a long period of time, especially the proportion of sweet potato weevils at different growth stages. The analysis here is based on the ratio of adults to larvae. Based on the analysis needs, the proportion analysis can be based on multiple growth stages. The data dimensions and parameters of the first and second evaluation data are consistent. Each node corresponds to independent sampling data and analysis process. The first and second evaluation data include pest assessment information under each node.

[0062] According to an embodiment of the present invention, the S102 is specifically:

[0063] The image of the sampling point is acquired by the monitoring unit to obtain the sweet potato image data;

[0064] The sweet potato image data is preprocessed by image smoothing and noise reduction, and the sweet potato parts are segmented by edge detection operators to obtain multiple sweet potato part images;

[0065] Obtain comparative images of various parts of sweet potatoes under different growth conditions of sweet potato weevils from the comparison database. Unify the image size and color space of the sweet potato partial images and the corresponding comparison images. Set a 10x10 pixel window and calculate the similarity of each window in the form of SSIM by sliding the window. Finally, complete the calculation of the entire image area and obtain the SSIM mean.

[0066] The similarity between some sweet potato images and comparison images was analyzed by SSIM mean, and the pest status of sweet potato elephants was evaluated based on the similarity analysis of multiple comparison images;

[0067] SSIM similarity analysis and comparison were performed on multiple partial images of sweet potatoes, and a comprehensive pest assessment was conducted on the entire planting area based on the sampling points to generate the second assessment data.

[0068] It should be noted that the monitoring unit includes a camera device for capturing images and performing pest analysis on various parts of the sweet potato. The analysis of various parts of the sweet potato includes vines, leaves, tender shoots, and tuberous roots, which are easily affected by pests. SSIM is the structural similarity index. In one part of the sweet potato, the comparison images include multiple types, that is, comparison images under multiple pest conditions. Specifically, multiple comparison images under different adult and larval ratios are compared with the sweet potato part image for similarity. Through the comparison results, the pest status of the current sweet potato image is analyzed based on the high similarity value, and evaluation data is generated. LSTM is a long short-term memory network. Based on general analysis needs, sequence fitting analysis methods such as linear regression prediction can be used here.

[0069] The edge detection operator may use detection operators such as Sobel to extract edge features.

[0070] According to an embodiment of the present invention, the S103 is specifically as follows:

[0071] Performing adult-larva ratio analysis and numerical extraction based on the first evaluation data and the second evaluation data to obtain a numerical ratio for each node;

[0072] Based on the first evaluation data, serializing the proportional values ​​of N consecutive nodes to obtain a sequence segment, and generating multiple sequence segments based on multiple nodes;

[0073] generating a corresponding plurality of sequence segments based on the second evaluation data;

[0074] The sequence segments corresponding to the first and second evaluation data are used as monitoring growth sequences, and each sequence segment is vectorized to generate multiple feature vectors;

[0075] Build a GAN-based generative model, which includes a generator, a discriminator, a loss function, and an optimizer;

[0076] Import multiple feature vectors as real data into the generator for self-learning, and generate simulated features through the generator;

[0077] In the discriminator, the loss function is set based on the mean square error between the simulated features and the real data, and the accuracy of the simulated features is determined;

[0078] The discriminator and the generator are trained in an adversarial manner until they reach Nash equilibrium. The convergence value of the loss function and the optimizer status at this time are recorded to complete the model training.

[0079] It should be noted that, in the proportional value of each node obtained, the proportional value of each node obtained by the first evaluation data and the proportional value of each node obtained by the second evaluation data are included. N is a user-set value, for example, can be set to 3, i.e., the proportional values ​​of 3 consecutive nodes form a sequence segment, and the vector obtained, i.e., the eigenvector with a dimension of 3, if there are 180 nodes in total, then 60 sequence segments are generated. The monitoring growth sequence reflects the growth status and proportional situation of the sweet potato weevil at different nodes. The present invention carries out simulation analysis and sequence prediction in the form of sequence, and can realize the efficient and accurate prediction and evaluation of the growth situation of the sweet potato weevil in the planting area.

[0080] It can be understood here that due to the difference in evaluation methods between the first and second evaluation data, when planting sweet potatoes in multiple regions or complex scenarios, the corresponding growth sequences in the first and second evaluation data can be weighted averaged and used as real data for model training. For example, for the proportion value under a certain node, the proportions corresponding to the first and second evaluation data are K1 and K2. K1 and K2 are weighted averaged to obtain a new proportion value, and multiple nodes (or all nodes) are selected for weighted averaging to obtain a weighted growth sequence, which is finally applied to model training.

[0081] According to an embodiment of the present invention, the S104 is specifically as follows:

[0082] During a real-time monitoring cycle, we conducted an assessment of sweet potato weevil pests based on multiple sampling points, analyzed the ratio of adults to larvae at multiple nodes, and serialized the ratio values ​​to obtain a real-time growth sequence.

[0083] The real-time growth sequence is imported into the GAN model for sequence feature learning, and simulated data is generated based on the generator cycle. The discriminator extracts the simulated data with a probability greater than the preset discrimination probability to obtain the optimal simulated data. The simulation generation is repeated until the preset amount of optimal simulated data is obtained.

[0084] The optimized simulation data is subjected to data analysis to obtain a simulation sequence.

[0085] It should be noted that during the real-time monitoring cycle, based on actual monitoring resources and needs, sweet potato sampling or image analysis techniques can be used to analyze the ratio of adults to larvae and generate real-time growth sequences. Data with a probability greater than the preset discrimination probability are selected as having a higher than expected accuracy. Simulated sequences include multiple segments, all derived from simulated analysis of real sequences.

[0086] According to an embodiment of the present invention, the S105 is specifically as follows:

[0087] Sequentially combine the real-time growth sequence and the simulation sequence to obtain a combined sequence;

[0088] The LSTM network is used to analyze and predict the sequence characteristics of the combined sequence, set the preset time nodes, and obtain the growth prediction sequence;

[0089] Data analysis of growth prediction sequences was performed to evaluate the growth trends and stages of the sweet potato weevil, predict pest distribution and set pest early warning information, and dynamically set prevention and control plans and monitoring plans based on the ratio of adults to larvae.

[0090] It should be noted that sequence combinations can be combined by combining a real-time growth sequence with a certain segment of the simulation sequence, and based on these combinations, multiple combination sequences can be formed to simulate the growth of pests in a planting area under complex conditions. Control plans include microbial agents for adults or larvae, sex attractants, light trapping, color plate trapping, and setting egg mass removal cycles and crop rotation cycles. Dynamic concentrations of agents and pesticides can be set for different adult-larvae ratios. Growth prediction sequences can also be used to predict the growth trend of the sweet potato weevil and adjust the monitoring plan to suit different growth stages and conditions, effectively allocating monitoring resources.

[0091] It is worth mentioning here that in large-scale or multi-crop sweet potato planting environments, traditional technologies are often difficult to achieve accurate and efficient analysis and prediction of the impact of sweet potato weevils, and generally rely on manual experience. For existing monitoring methods, the amount of monitoring data for sweet potato weevils is often low, making it difficult to conduct long-term and efficient prediction analysis, and it is difficult to simulate the growth trend of sweet potato weevils in complex environments. In addition, traditional pest control for sweet potato weevils is often based on repeated single-type control plans, such as long-term control with a single pesticide plan, which can easily lead to drug resistance or certain adaptability of sweet potato pests in the field. It lacks the ability to analyze dynamic pests and set plans, relies heavily on manual experience analysis and control, is not very practical, and makes it difficult for sweet potato planting to develop at a high level.

[0092] Based on this, the present invention analyzes and calculates pest assessment data of historical periods through sampling and image methods, and sets the monitoring growth sequence of the sweet potato weevil. The GAN model is trained based on the existing monitoring growth sequence to simulate the growth of pests in the target planting environment in sequence form. The GAN model is used to analyze the growth sequence law of the sweet potato weevil in the actual sweet potato area, and perform sequence simulation through the generator and discriminator. It can simulate the growth sequence of pests with less monitoring sequence data, and combine the simulated data with the real-time data, and perform prediction through the LSTM network to analyze the potential growth trend of the sweet potato weevil, which is effectively applied to the growth situation and growth stage prediction of the sweet potato weevil in multiple planting scenarios.

[0093] Through the present invention, highly accurate predictive analysis and current pest growth status prediction and evaluation can be carried out, thereby improving the accuracy of pest assessment in planting areas with fewer monitoring points or fewer monitoring units, further providing scientific decision-making for the prevention and control of sweet potato weevils in multiple planting scenarios, and achieving a higher level of comprehensive analysis and management.

[0094] According to an embodiment of the present invention, the further embodiment includes:

[0095] Get the K real-time growth sequences corresponding to the latest K real-time monitoring cycles. The number of nodes corresponding to each real-time growth sequence is P.

[0096] For the generative model of the current state, set the number of prediction nodes to P and generate K real-time simulation sequences;

[0097] Extract a sequence from each of the K real-time growth sequences and the K real-time simulation sequences, and mark them as a reference sequence and a comparison sequence;

[0098] The resolution coefficient is set to 0.5, and the correlation between the reference series and the comparison series is calculated based on the grey correlation method;

[0099] If the correlation is lower than the preset correlation, a certain number of sequences are extracted from the K real-time growing sequences as training sequence data based on the correlation;

[0100] The training sequence data is imported into the generative model for secondary training and parameter setting to obtain the generative model after secondary training.

[0101] It should be noted that, based on the changes in the planting environment and the impact of pests, if there is a large-scale mutation in the number of pests or a large change in the planting environment conditions, it may cause the simulation sequence of the generated model to deviate greatly from the growth sequence of the real-time period, and the corresponding simulation data to deviate greatly, which may cause the simulation sequence to be unreliable for reference. Therefore, the present invention uses the grey correlation method to perform a combined correlation evaluation on the corresponding real-time sequences. If the correlation is low, the training data is dynamically set to optimize the model, which can enable the model to be dynamically adjusted according to the planting environment to improve practicality.

[0102] Based on the correlation, a certain number of sequences are extracted from K real-time growth sequences as training sequence data. The lower the correlation, the more sequences are extracted. This extracted data is used to optimize the model parameters to simulate sequence analysis in a realistic planting environment. The preset correlation can be set in the range of [0.5, 0.9]. When extracting sequences and marking reference and comparison sequences, a total of K×K reference and comparison sequences can be extracted for correlation analysis. Correlation analysis is not performed on repeated sequence combinations.

[0103] Figure 2 A block diagram of a sweet potato weevil impact distribution analysis platform based on simulation analysis of the present invention is shown.

[0104] A second aspect of the present invention further provides a sweet potato weevil impact distribution analysis platform 2 based on simulation analysis, the platform comprising: a memory 21 and a processor 22, wherein the memory 21 includes a sweet potato weevil impact distribution analysis program based on simulation analysis, and when the sweet potato weevil impact distribution analysis program based on simulation analysis is executed by the processor 22, the following steps are implemented:

[0105] S101: In a historical monitoring period, multiple monitoring time nodes are set, multiple sampling points are set based on a target sweet potato planting area, and first evaluation data based on the sweet potato weevil is obtained at the sampling points;

[0106] S102: acquiring sweet potato image data based on the sampling points, performing SSIM image similarity analysis based on a comparison database, performing pest assessment based on the similarity, and obtaining second assessment data based on the sweet potato weevil;

[0107] S103: Analyzing the ratio of adults to larvae using the first and second evaluation data, and sorting the ratio values ​​based on multiple nodes to obtain a monitoring growth sequence, constructing a generative model based on a GAN, using the monitoring growth sequence as training data, and performing cyclic adversarial training on the generative model until the model reaches a Nash equilibrium;

[0108] S104: Based on the real-time monitoring cycle, the ratio of adults to larvae in the target sweet potato planting area is analyzed, and a real-time growth sequence is generated. The real-time growth sequence is introduced into the generative model for sequence learning and generation of a simulation sequence;

[0109] S105: Sequence combination is performed based on the real-time growth sequence and the simulation sequence, and sequence prediction is performed on the combined sequence through the LSTM network to obtain a growth prediction sequence based on the sweet potato weevil, and data analysis is performed on the growth prediction sequence to set a corresponding prevention and control plan.

[0110] It should be noted that, in the present invention, a node refers to a monitoring time node.

[0111] According to an embodiment of the present invention, the S101 is specifically as follows:

[0112] Based on a historical monitoring period, multiple monitoring time nodes are set to ensure that the time intervals between two adjacent nodes are the same;

[0113] Based on the size of the target sweet potato area, multiple sampling points are set, and multiple sweet potato plants are sampled and pests are assessed at the sampling points to obtain first assessment data;

[0114] Pest assessment includes recording information on the number of adults, larvae, developmental stages, and damage symptoms of the sweet potato weevil.

[0115] It should be noted that the historical monitoring period is a longer monitoring cycle, which is used to analyze the growth patterns of sweet potato weevils in the planting area over a long period of time, especially the proportion of sweet potato weevils at different growth stages. The analysis here is based on the ratio of adults to larvae. Based on the analysis needs, the proportion analysis can be based on multiple growth stages. The data dimensions and parameters of the first and second evaluation data are consistent. Each node corresponds to independent sampling data and analysis process. The first and second evaluation data include pest assessment information under each node.

[0116] According to an embodiment of the present invention, the S102 is specifically:

[0117] The image of the sampling point is acquired by the monitoring unit to obtain the sweet potato image data;

[0118] The sweet potato image data is preprocessed by image smoothing and noise reduction, and the sweet potato parts are segmented by edge detection operators to obtain multiple sweet potato part images;

[0119] Obtain comparative images of various parts of sweet potatoes under different growth conditions of sweet potato weevils from the comparison database. Unify the image size and color space of the sweet potato partial images and the corresponding comparison images. Set a 10x10 pixel window and calculate the similarity of each window in the form of SSIM by sliding the window. Finally, complete the calculation of the entire image area and obtain the SSIM mean.

[0120] The similarity between some sweet potato images and comparison images was analyzed by SSIM mean, and the pest status of sweet potato elephants was evaluated based on the similarity analysis of multiple comparison images;

[0121] SSIM similarity analysis and comparison were performed on multiple partial images of sweet potatoes, and a comprehensive pest assessment was conducted on the entire planting area based on the sampling points to generate the second assessment data.

[0122] It should be noted that the monitoring unit includes a camera device for capturing images and performing pest analysis on various parts of the sweet potato. The analysis of various parts of the sweet potato includes vines, leaves, tender shoots, and tuberous roots, which are easily affected by pests. SSIM is the structural similarity index. In one part of the sweet potato, the comparison images include multiple types, that is, comparison images under multiple pest conditions. Specifically, multiple comparison images under different adult and larval ratios are compared with the sweet potato part image for similarity. Through the comparison results, the pest status of the current sweet potato image is analyzed based on the high similarity value, and evaluation data is generated. LSTM is a long short-term memory network. Based on general analysis needs, sequence fitting analysis methods such as linear regression prediction can be used here.

[0123] The edge detection operator may use detection operators such as Sobel to extract edge features.

[0124] According to an embodiment of the present invention, the S103 is specifically as follows:

[0125] Performing adult-larva ratio analysis and numerical extraction based on the first evaluation data and the second evaluation data to obtain a numerical ratio for each node;

[0126] Based on the first evaluation data, serializing the proportional values ​​of N consecutive nodes to obtain a sequence segment, and generating multiple sequence segments based on multiple nodes;

[0127] generating a corresponding plurality of sequence segments based on the second evaluation data;

[0128] The sequence segments corresponding to the first and second evaluation data are used as monitoring growth sequences, and each sequence segment is vectorized to generate multiple feature vectors;

[0129] Build a GAN-based generative model, which includes a generator, a discriminator, a loss function, and an optimizer;

[0130] Import multiple feature vectors as real data into the generator for self-learning, and generate simulated features through the generator;

[0131] In the discriminator, the loss function is set based on the mean square error between the simulated features and the real data, and the accuracy of the simulated features is determined;

[0132] The discriminator and the generator are trained in an adversarial manner until they reach Nash equilibrium. The convergence value of the loss function and the optimizer status at this time are recorded to complete the model training.

[0133] It should be noted that, in the proportional value of each node obtained, the proportional value of each node obtained by the first evaluation data and the proportional value of each node obtained by the second evaluation data are included. N is a user-set value, for example, can be set to 3, i.e., the proportional values ​​of 3 consecutive nodes form a sequence segment, and the vector obtained, i.e., the eigenvector with a dimension of 3, if there are 180 nodes in total, then 60 sequence segments are generated. The monitoring growth sequence reflects the growth status and proportional situation of the sweet potato weevil at different nodes. The present invention carries out simulation analysis and sequence prediction in the form of sequence, and can realize the efficient and accurate prediction and evaluation of the growth situation of the sweet potato weevil in the planting area.

[0134] It can be understood here that due to the difference in evaluation methods between the first and second evaluation data, when planting sweet potatoes in multiple regions or complex scenarios, the corresponding growth sequences in the first and second evaluation data can be weighted averaged and used as real data for model training. For example, for the proportion value under a certain node, the proportions corresponding to the first and second evaluation data are K1 and K2. K1 and K2 are weighted averaged to obtain a new proportion value, and multiple nodes (or all nodes) are selected for weighted averaging to obtain a weighted growth sequence, which is finally applied to model training.

[0135] According to an embodiment of the present invention, the S104 is specifically as follows:

[0136] During a real-time monitoring cycle, we conducted an assessment of sweet potato weevil pests based on multiple sampling points, analyzed the ratio of adults to larvae at multiple nodes, and serialized the ratio values ​​to obtain a real-time growth sequence.

[0137] The real-time growth sequence is imported into the GAN model for sequence feature learning, and simulated data is generated based on the generator cycle. The discriminator extracts the simulated data with a probability greater than the preset discrimination probability to obtain the optimal simulated data. The simulation generation is repeated until the preset amount of optimal simulated data is obtained.

[0138] The optimized simulation data is subjected to data analysis to obtain a simulation sequence.

[0139] It should be noted that during the real-time monitoring cycle, based on actual monitoring resources and needs, sweet potato sampling or image analysis techniques can be used to analyze the ratio of adults to larvae and generate real-time growth sequences. Data with a probability greater than the preset discrimination probability are selected as having a higher than expected accuracy. Simulated sequences include multiple segments, all derived from simulated analysis of real sequences.

[0140] According to an embodiment of the present invention, the S105 is specifically as follows:

[0141] Sequentially combine the real-time growth sequence and the simulation sequence to obtain a combined sequence;

[0142] The LSTM network is used to analyze and predict the sequence characteristics of the combined sequence, set the preset time nodes, and obtain the growth prediction sequence;

[0143] Data analysis of growth prediction sequences was performed to evaluate the growth trends and stages of the sweet potato weevil, predict pest distribution and set pest early warning information, and dynamically set prevention and control plans and monitoring plans based on the ratio of adults to larvae.

[0144] It should be noted that sequence combinations can be combined by combining a real-time growth sequence with a certain segment of the simulation sequence, and based on these combinations, multiple combination sequences can be formed to simulate the growth of pests in a planting area under complex conditions. Control plans include microbial agents for adults or larvae, sex attractants, light trapping, color plate trapping, and setting egg mass removal cycles and crop rotation cycles. Dynamic concentrations of agents and pesticides can be set for different adult-larvae ratios. Growth prediction sequences can also be used to predict the growth trend of the sweet potato weevil and adjust the monitoring plan to suit different growth stages and conditions, effectively allocating monitoring resources.

[0145] It is worth mentioning here that in large-scale or multi-crop sweet potato planting environments, traditional technologies are often difficult to achieve accurate and efficient analysis and prediction of the impact of sweet potato weevils, and generally rely on manual experience. For existing monitoring methods, the amount of monitoring data for sweet potato weevils is often low, making it difficult to conduct long-term and efficient prediction analysis, and it is difficult to simulate the growth trend of sweet potato weevils in complex environments. In addition, traditional pest control for sweet potato weevils is often based on repeated single-type control plans, such as long-term control with a single pesticide plan, which can easily lead to drug resistance or certain adaptability of sweet potato pests in the field. It lacks the ability to analyze dynamic pests and set plans, relies heavily on manual experience analysis and control, is not very practical, and makes it difficult for sweet potato planting to develop at a high level.

[0146] Based on this, the present invention analyzes and calculates pest assessment data of historical periods through sampling and image methods, and sets the monitoring growth sequence of the sweet potato weevil. The GAN model is trained based on the existing monitoring growth sequence to simulate the growth of pests in the target planting environment in sequence form. The GAN model is used to analyze the growth sequence law of the sweet potato weevil in the actual sweet potato area, and perform sequence simulation through the generator and discriminator. It can simulate the growth sequence of pests with less monitoring sequence data, and combine the simulated data with the real-time data, and perform prediction through the LSTM network to analyze the potential growth trend of the sweet potato weevil, which is effectively applied to the growth situation and growth stage prediction of the sweet potato weevil in multiple planting scenarios.

[0147] Through the present invention, highly accurate predictive analysis and current pest growth status prediction and evaluation can be carried out, thereby improving the accuracy of pest assessment in planting areas with fewer monitoring points or fewer monitoring units, further providing scientific decision-making for the prevention and control of sweet potato weevils in multiple planting scenarios, and achieving a higher level of comprehensive analysis and management.

[0148] The third aspect of the present invention also provides a computer-readable storage medium, which includes a sweet potato weevil impact distribution analysis program based on simulation analysis. When the sweet potato weevil impact distribution analysis program based on simulation analysis is executed by a processor, it implements the steps of the sweet potato weevil impact distribution analysis method based on simulation analysis as described in any one of the above items.

[0149] The present invention discloses a sweet potato weevil impact distribution analysis platform based on simulation analysis, comprising: setting multiple time nodes and sampling points within a historical monitoring period and analyzing first assessment data of the sweet potato weevil; combining sweet potato image data with a comparison database, analyzing pest status through SSIM similarity, and obtaining second assessment data; integrating the two types of data to generate a monitoring growth sequence based on pest proportion, performing feature learning on the growth sequence and training a GAN generation model; collecting real-time growth sequences, inputting them into a generation model to obtain simulation sequences, combining the real-time growth sequences with the simulation sequences, using an LSTM network to predict the growth trend of the sweet potato weevil, and dynamically formulating targeted prevention and control plans and monitoring plans. The present invention can dynamically predict and analyze the growth status and number changes of the sweet potato weevil, effectively improving the efficiency of sweet potato pest monitoring and control.

[0150] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0151] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0152] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0153] Those skilled in the art will appreciate that all or part of the steps of the above-mentioned method embodiments may be implemented by hardware associated with program instructions, and the aforementioned program may be stored in a computer-readable storage medium. When the program is executed, the program executes the steps of the above-mentioned method embodiments. The aforementioned storage medium includes various media that can store program codes, such as mobile storage devices, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0154] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0155] The above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis, characterized in that: include: S101: In a historical monitoring period, multiple monitoring time nodes are set, multiple sampling points are set based on a target sweet potato planting area, and first evaluation data based on the sweet potato weevil is obtained at the sampling points; S102: acquiring sweet potato image data based on the sampling points, performing SSIM image similarity analysis based on a comparison database, performing pest assessment based on the similarity, and obtaining second assessment data based on the sweet potato weevil; S103: Analyzing the ratio of adults to larvae using the first and second evaluation data, and sorting the ratio values ​​based on multiple nodes to obtain a monitoring growth sequence, constructing a generative model based on a GAN, using the monitoring growth sequence as training data, and performing cyclic adversarial training on the generative model until the model reaches a Nash equilibrium; S104: Based on the real-time monitoring cycle, the ratio of adults to larvae in the target sweet potato planting area is analyzed, and a real-time growth sequence is generated. The real-time growth sequence is introduced into the generative model for sequence learning and generation of a simulation sequence; S105: Sequence combination is performed based on the real-time growth sequence and the simulation sequence, and sequence prediction is performed on the combined sequence through the LSTM network to obtain a growth prediction sequence based on the sweet potato weevil, and data analysis is performed on the growth prediction sequence to set a corresponding prevention and control plan.

2. The method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis according to claim 1, wherein: The S101 is specifically as follows: Based on a historical monitoring period, multiple monitoring time nodes are set to ensure that the time intervals between two adjacent nodes are the same; Based on the size of the target sweet potato area, multiple sampling points are set, and multiple sweet potato plants are sampled and pests are assessed at the sampling points to obtain first assessment data; Pest assessment includes recording information on the number of adults, larvae, developmental stages, and damage symptoms of the sweet potato weevil.

3. The method for analyzing the impact distribution of sweet potato weevils based on simulation analysis according to claim 1, wherein: The S102 is specifically as follows: The image of the sampling point is acquired by the monitoring unit to obtain the sweet potato image data; The sweet potato image data is preprocessed by image smoothing and noise reduction, and the sweet potato parts are segmented by edge detection operators to obtain multiple sweet potato part images; Obtain comparative images of various parts of sweet potatoes under different growth conditions of sweet potato weevils from the comparison database. Unify the image size and color space of the sweet potato partial images and the corresponding comparison images. Set a 10x10 pixel window and calculate the similarity of each window in the form of SSIM by sliding the window. Finally, complete the calculation of the entire image area and obtain the SSIM mean. The similarity between some sweet potato images and comparison images was analyzed by SSIM mean, and the pest status of sweet potato elephants was evaluated based on the similarity analysis of multiple comparison images; SSIM similarity analysis and comparison were performed on multiple partial images of sweet potatoes, and a comprehensive pest assessment was conducted on the entire planting area based on the sampling points to generate the second assessment data.

4. The method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis according to claim 1, wherein: The S103 is specifically as follows: Performing adult-larva ratio analysis and numerical extraction based on the first evaluation data and the second evaluation data to obtain a numerical ratio for each node; Based on the first evaluation data, serializing the proportional values ​​of N consecutive nodes to obtain a sequence segment, and generating multiple sequence segments based on multiple nodes; generating a corresponding plurality of sequence segments based on the second evaluation data; The sequence segments corresponding to the first and second evaluation data are used as monitoring growth sequences, and each sequence segment is vectorized to generate multiple feature vectors; Build a GAN-based generative model, which includes a generator, a discriminator, a loss function, and an optimizer; Import multiple feature vectors as real data into the generator for self-learning, and generate simulated features through the generator; In the discriminator, the loss function is set based on the mean square error between the simulated features and the real data, and the accuracy of the simulated features is determined; The discriminator and the generator are trained in an adversarial manner until they reach Nash equilibrium. The convergence value of the loss function and the optimizer status at this time are recorded to complete the model training.

5. The method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis according to claim 1, characterized in that: The S104 is specifically as follows: During a real-time monitoring cycle, we conducted an assessment of sweet potato weevil pests based on multiple sampling points, analyzed the ratio of adults to larvae at multiple nodes, and serialized the ratio values ​​to obtain a real-time growth sequence. The real-time growth sequence is imported into the GAN model for sequence feature learning, and simulated data is generated based on the generator cycle. The discriminator extracts the simulated data with a probability greater than the preset discrimination probability to obtain the optimal simulated data. The simulation generation is repeated until the preset amount of optimal simulated data is obtained. The optimized simulation data is subjected to data analysis to obtain a simulation sequence.

6. The method for analyzing the distribution of the impact of sweet potato weevils based on simulation analysis according to claim 1, characterized in that: The S105 is specifically as follows: Sequentially combine the real-time growth sequence and the simulation sequence to obtain a combined sequence; The LSTM network is used to analyze and predict the sequence characteristics of the combined sequence, set the preset time nodes, and obtain the growth prediction sequence; Data analysis of growth prediction sequences was performed to evaluate the growth trends and stages of the sweet potato weevil, predict pest distribution and set pest early warning information, and dynamically set prevention and control plans and monitoring plans based on the ratio of adults to larvae.

7. A platform for analyzing the impact distribution of sweet potato weevils based on simulation analysis, characterized in that: The platform includes: a memory and a processor. The memory includes a sweet potato weevil impact distribution analysis program based on simulation analysis. When the sweet potato weevil impact distribution analysis program based on simulation analysis is executed by the processor, the following steps are implemented: S101: In a historical monitoring period, multiple monitoring time nodes are set, multiple sampling points are set based on a target sweet potato planting area, and first evaluation data based on the sweet potato weevil is obtained at the sampling points; S102: acquiring sweet potato image data based on the sampling points, performing SSIM image similarity analysis based on a comparison database, performing pest assessment based on the similarity, and obtaining second assessment data based on the sweet potato weevil; S103: Analyzing the ratio of adults to larvae using the first and second evaluation data, and sorting the ratio values ​​based on multiple nodes to obtain a monitoring growth sequence, constructing a generative model based on a GAN, using the monitoring growth sequence as training data, and performing cyclic adversarial training on the generative model until the model reaches a Nash equilibrium; S104: Based on the real-time monitoring cycle, the ratio of adults to larvae in the target sweet potato planting area is analyzed, and a real-time growth sequence is generated. The real-time growth sequence is introduced into the generative model for sequence learning and generation of a simulation sequence; S105: Sequence combination is performed based on the real-time growth sequence and the simulation sequence, and sequence prediction is performed on the combined sequence through the LSTM network to obtain a growth prediction sequence based on the sweet potato weevil, and data analysis is performed on the growth prediction sequence to set a corresponding prevention and control plan.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a sweet potato weevil impact distribution analysis program based on simulation analysis. When the sweet potato weevil impact distribution analysis program based on simulation analysis is executed by a processor, the steps of the sweet potato weevil impact distribution analysis method based on simulation analysis as described in any one of claims 1 to 6 are implemented.

Images