Soybean cyst nematode population dynamic monitoring method and system based on the Internet of Things

Abstract

The present invention discloses a method and system for monitoring soybean cyst nematode population dynamics based on the Internet of Things (IoT). The method comprises collecting pest detection data and monitoring data within a preset area for gene-driven pest control, preprocessing the pest detection data and monitoring data, performing target screening on the pest detection data to obtain target detection data, and performing control risk assessment on the target detection data based on the monitoring data to obtain a control ratio, performing causal analysis on the pest detection data based on gene drive to obtain an ecological risk, constructing an optimized growth degree function based on the control ratio and the ecological risk, constructing an IoT-based soybean cyst nematode population dynamics monitoring model based on the optimized growth degree function, inputting the data to be monitored into the IoT-based soybean cyst nematode population dynamics monitoring model, and outputting monitoring results. This method improves the accuracy of nematode population dynamics monitoring and can be directly applied to nematode population dynamics monitoring systems.

Description

Technical Field

[0001] The present invention relates to the field of dynamic monitoring, and in particular to a method and system for dynamic monitoring of soybean cyst nematode populations based on the Internet of Things. Background Art

[0002] Gene drive technology provides a revolutionary means for regional pest control by regulating the genes of pest populations in a targeted manner. However, the dynamic balance between its ecological risks and control effectiveness remains a core challenge. Traditional monitoring methods rely on static statistical models or single environmental parameter analysis, making it difficult to cope with the complex ecological chain reactions triggered by gene drive technology. For example, existing technologies lack the ability to quantitatively assess the coupled effects of gene mutation transmission, population dynamics, and ecosystem feedback. In addition, there are bottlenecks in the efficient integration and real-time analysis of high-dimensional heterogeneous data, resulting in delayed risk assessments and insufficiently accurate prevention and control decisions.

[0003] With breakthroughs in intelligent algorithms, deep learning, causal reasoning, and multi-objective optimization have provided new approaches for dynamic monitoring. The core risk of gene drive technology lies in its self-propagating nature, which can trigger irreversible ecological impacts. This requires the establishment of a real-time assessment system encompassing multiple dimensions, including genetic variation, population dynamics, and environmental responses. However, current research still faces three major limitations: The high dimensionality and significant spatiotemporal heterogeneity of monitoring data make it difficult to identify key risk factors using traditional feature screening methods; the complex causal relationship between gene drive interventions and ecological risks makes static models unable to capture dynamic evolutionary patterns; and the optimization of prevention and control strategies lacks a mechanism for balancing the dual objectives of "short-term efficiency and long-term safety."

[0004] Therefore, there is an urgent need to build a dynamic monitoring framework based on intelligent algorithms to reduce the monitoring lag and risk control problems of gene drive technology, and provide technical support for precise prevention and control and ecological security. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for monitoring soybean cyst nematode population dynamics based on the Internet of Things.

[0006] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0007] The present invention comprises the following steps:

[0008] Collecting pest detection data and monitoring data within a preset area for gene-driven pest control, and pre-processing the pest detection data and monitoring data; the monitoring data includes historical data and data to be analyzed; the data to be analyzed includes pest data, environmental data, and ecosystem data; the pest detection data includes species, quantity, distribution location, infection level, growth stage of the pest, and gene sequence; the historical data includes control measures and historical monitoring data;

[0009] Performing target screening on the pest detection data to obtain target detection data, and performing prevention and control risk assessment on the target detection data based on the monitoring data to obtain a prevention and control ratio;

[0010] Performing causal analysis on the pest detection data according to gene drive to obtain an ecological risk degree, and constructing an optimized growth degree function according to the control ratio and the ecological risk degree;

[0011] An Internet of Things soybean cyst nematode population dynamic monitoring model is constructed according to the optimized growth degree function, the data to be monitored is input into the Internet of Things soybean cyst nematode population dynamic monitoring model, and the monitoring results are output.

[0012] Furthermore, the method of performing target screening on the pest detection data to obtain target detection data includes:

[0013] The pest growth stage and the degree of crop damage and infection are obtained based on pest detection data. The pest development status is divided according to the number of pests at different growth stages. The pest danger level is obtained based on the degree of crop damage and infection. Pest data at key growth stages with high infection levels are prioritized, and pest individuals that are only in the egg stage or have a damage and infection level below 0.132 are eliminated to obtain primary data.

[0014] The gene drive matching degree is calculated based on the homology between the primary data and the target gene based on the pest gene sequence, and the primary data with a gene drive matching degree greater than the matching degree threshold is used as the key data;

[0015] Extract key features of key data, calculate the importance of key features using the random forest algorithm, sort key features in descending order based on importance, dynamically adjust the screening strategy based on drug resistance, and prioritize the key features with the highest scores;

[0016] Spatial autocorrelation analysis was performed on key data to identify pest clustering areas and obtain pest distribution hotspots. Spatiotemporal scanning statistical methods were used to detect specific time periods with drastic changes in pests. Dissimilarity detection was performed on key data within specific time periods, and key data with a dissimilarity higher than 0.751 were output as target detection data.

[0017] Furthermore, the method of performing a prevention and control risk assessment on the target detection data based on the monitoring data to obtain a prevention and control ratio includes:

[0018] Extract features from monitoring data to obtain monitoring features, select monitoring features based on spatiotemporal features to obtain spatiotemporal features;

[0019] Classification features are obtained by classifying target detection data, and interaction features are generated by matching spatiotemporal features with classification features. Classification features include pest characteristics, environmental characteristics, and gene drive association features. Pest characteristics include reproduction rate, spread speed, pesticide resistance index, and growth stage weight. Environmental characteristics include temperature and humidity adaptability and host plant coverage. Gene drive association features include target gene editing efficiency and gene drive spread potential.

[0020] Use ensemble learning methods to combine random forest and quarterly gradient boosting trees to build a prevention and control risk assessment model;

[0021] Calculate the degree of harm, controllability and ecological sensitivity based on the interaction characteristics:

[0022]

[0023]

[0024]

[0025] The hazard level is , the controllability is , the ecological sensitivity is , the weight coefficients are 、 、 、 、 , the reproduction rate is , the diffusion rate is The infection level is , the gene drive matching degree is , the target gene editing efficiency is , the drug resistance index is, and the smoothing factor is , the host plant coverage is , the temperature and humidity difference is ;

[0026] Calculate the control ratio:

[0027]

[0028] The prevention ratio is , the risk weight is , the sensitive weight is ;

[0029] The learning rate of the prevention and control risk assessment model is optimized through grid search, the target detection data is input into the optimized prevention and control risk assessment model, and the prevention and control ratio is output.

[0030] Furthermore, the method of performing causal analysis on the pest detection data based on gene drive to obtain ecological risk includes:

[0031] Obtain ecosystem data based on monitoring data, including abundance, population structure, food web interactions, and environmental parameters of non-target species;

[0032] Intervention variables are obtained based on the actual release rate of gene drive and the target pest population, and outcome variables are obtained based on the change rate of the target pest population and the changes in the abundance of non-target associated species; confounding variables are obtained based on environmental data and current control measures.

[0033] Confounding variables and mediating variables are added to refine the causal relationship. A causal graph based on a Bayesian network is constructed based on the refined causal relationship. When randomized controlled experiments are not possible, instrumental variables are introduced to perform anchored causal analysis to obtain enhanced causal relationships. A graph neural network is introduced to process the complex ecological network relationship based on the enhanced causal relationship to obtain the ecological causal relationship. The ecological network relationship is the causal relationship between pests, gene drives, and ecological functions. The mediating variable is the food web interaction relationship. The expression of the causal graph is:

[0034]

[0035] The outcome variable is Ot, the treatment variable is Tr, the confounding variable is Cf, and the marginal probability of the outcome variable is , the marginal probability of the treatment variable is , the conditional probability of the treatment variable under the confounding variable Cf is , the conditional probability distribution under the conditions of treatment variable Tr and confounding variable Cf is ;

[0036] Use the Bayesian formula to update the posterior distribution, the expression is:

[0037]

[0038] The Bayesian network parameters are , the new pest detection data is , the historical pest detection data is , Bayesian network parameters Download new pest detection data The probability of generating , new pest detection data Conditional Bayesian network parameters The posterior distribution of , historical pest detection data Conditional Bayesian network parameters The posterior distribution of ;

[0039] Calculate ecological risk based on objective weighting:

[0040]

[0041] The ecological risk is , the target pest reduction rate is The expected decline rate is , the non-target species impact index is , the ecosystem health index is , the ecological health weight is , the weight of non-target species is , the decline rate weight is ;

[0042] Through machine learning, the ecological health weight, non-target species weight, and decline rate weight are dynamically adjusted, and the pest detection data is input into the adjusted causal diagram to output the ecological risk level.

[0043] Furthermore, a method for constructing an optimized growth degree function based on the control ratio and the ecological risk degree includes:

[0044] A directed weight mechanism is introduced to construct the control directed weight and ecological directed weight according to the control ratio and ecological risk. The expression is:

[0045]

[0046]

[0047] The prevention and control directed weight is , the ecological directed weight is , the prevention and treatment ratio at time t is The historical average prevention and control ratio is The standard deviation of the control ratio is , the ecological risk at time t is The safety threshold of ecological risk is The standard deviation of ecological risk is , the basic weight of the control ratio is , the basic weight of ecological risk is , the sensitivity coefficient is , the hyperbolic tangent function is ;

[0048] when When it is greater than zero, the control is effective; when When it is less than zero, the prevention and treatment is ineffective and the condition worsens; when When it is greater than zero, the ecological risk exceeds the safety range; when When it is less than , the ecological risk is controllable;

[0049] The optimal growth degree function is constructed based on the prevention and control directed weights and the ecological directed weights, and the expression is:

[0050]

[0051] The optimized growth function at time t is , the normalized prevention and control ratio at time t is , the normalized ecological risk at time t is .

[0052] Furthermore, a method for constructing an IoT soybean cyst nematode population dynamic monitoring model based on the optimized growth degree function includes:

[0053] The objective weighted sum of the optimized growth degree function and the loss function is used as the objective function of the Internet of Things soybean cyst nematode population dynamics monitoring model.

[0054] The IoT-based soybean cyst nematode population dynamics monitoring model includes autoencoders, time series analysis algorithms, graph neural network algorithms, and multi-agent deep reinforcement learning;

[0055] The autoencoder encodes and compresses the input data into a low-dimensional representation, then decodes and reconstructs it. It automatically learns the intrinsic characteristics of the input data based on encoding and decoding to obtain the features to be analyzed.

[0056] Time series analysis algorithms build mathematical models to analyze the data variation patterns of the features to be analyzed at consecutive time points and capture time series characteristics. Time series characteristics include trends, seasonality, and periodicity.

[0057] The graph neural network algorithm converts time series characteristics into graph structure data, performs message passing and feature learning on the graph based on the objective function, captures the dynamic relationship of technological development, and realizes dynamic monitoring of priority development technologies;

[0058] Multi-agent deep reinforcement learning optimizes the Internet of Things soybean cyst nematode population dynamic monitoring model by having multiple agents interact and learn in the environment, and adjusting the strategy based on the feedback of the dynamic monitoring model.

[0059] Secondly, a dynamic monitoring system for priority development technologies based on intelligent algorithms, including:

[0060] Data acquisition module: used to collect pest detection data and monitoring data within the preset area of ​​gene-driven pest control, and pre-process the pest detection data and monitoring data; the monitoring data includes historical data and data to be analyzed; the data to be analyzed includes pest data, environmental data, and ecosystem data;

[0061] A screening risk assessment module is configured to perform target screening on the pest detection data to obtain target detection data, and perform control risk assessment on the target detection data based on the monitoring data to obtain a control ratio;

[0062] Causal fusion module: used to perform causal analysis on the pest detection data based on gene drive to obtain ecological risk, and construct an optimized growth degree function based on the control ratio and the ecological risk;

[0063] Modeling and optimization module: constructing an Internet of Things soybean cyst nematode population dynamic monitoring model according to the optimized growth degree function, inputting the data to be monitored into the Internet of Things soybean cyst nematode population dynamic monitoring model, and outputting the monitoring results.

[0064] The beneficial effects of the present invention are:

[0065] The present invention provides a method and system for monitoring soybean cyst nematode population dynamics based on the Internet of Things. Compared with the prior art, the present invention has the following technical effects:

[0066] The present invention can improve the accuracy of soybean cyst nematode population dynamic monitoring through preprocessing, target screening, prevention and control risk assessment, cause-and-effect analysis, construction and optimization of growth degree function and model construction steps, thereby improving the precision of soybean cyst nematode population dynamic monitoring. Optimizing soybean cyst nematode population dynamic monitoring can greatly save resources and improve work efficiency. It can realize soybean cyst nematode population dynamic monitoring, and perform prevention and control risk assessment and cause-and-effect analysis on soybean cyst nematode population dynamic monitoring in real time. It is of great significance to soybean cyst nematode population dynamic monitoring, can adapt to soybean cyst nematode population dynamic monitoring of different standards and different soybean cyst nematode population dynamic monitoring needs, and has a certain universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 The present invention is a flowchart of the steps of the method for dynamically monitoring soybean cyst nematode population based on the Internet of Things. DETAILED DESCRIPTION

[0068] The present invention will be further described below through specific examples. The illustrative examples and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0069] The method and system for monitoring soybean cyst nematode population dynamics based on the Internet of Things of the present invention include the following steps:

[0070] like Figure 1 As shown, in this embodiment, the following steps are included:

[0071] Collecting pest detection data and monitoring data within a preset area for gene-driven pest control, and pre-processing the pest detection data and monitoring data; the monitoring data includes historical data and data to be analyzed; the data to be analyzed includes pest data, environmental data, and ecosystem data; the pest detection data includes species, quantity, distribution location, infection level, growth stage of the pest, and gene sequence; the historical data includes control measures and historical monitoring data;

[0072] In actual assessments, ecosystem data include non-target species, relatively stable population structures, environmental parameters, and pest interactions;

[0073] The IoT sensor network collects pest detection data and environmental monitoring data from a preset area. The data is then transmitted to a cloud server for pre-processing via the LoRa wireless communication protocol. The IoT sensor network includes a soil temperature sensor, an image recognition camera, and a wireless data transmission module.

[0074] The study will examine the impact of soybean cyst nematodes on orchards where a gene-driven pest control system is used. Apples are the primary crop in these orchards. Monitoring data from January 2021 to April 2023 will be used as historical data.

[0075] IoT sensor nodes were deployed in a 50m x 50m grid across a 200-mu soybean planting area. These included: a soil sensor that collects soil moisture, temperature, and pH values ​​in real time; a 20-megapixel image recognition camera equipped with an AI pest recognition chip; a wireless transmission module using the LoRa protocol, with a transmission range of 3km and a power consumption of 10mW; and an edge computing gateway deployed in a regional center, responsible for data preprocessing and encryption.

[0076] The analysis period is from May to June 2023. Pest data: Reproduction rate 0.9, spread rate 0.7, insecticide resistance index 0.3, growth stage weight 0.8; Environmental data: Temperature and humidity adaptability 0.8, host plant coverage 0.7; Ecosystem data: Various insects and birds, soybean cyst nematodes are food for some birds and predatory insects, temperature 20-25°C, humidity 50-60%.

[0077] Control measures: Gene drive control will be adopted in June 2022, with a release of 10^ 5 infected individuals;

[0078] Historical monitoring data: 1.2 million soil and pest data stored via the IoT platform;

[0079] Performing target screening on the pest detection data to obtain target detection data, and performing prevention and control risk assessment on the target detection data based on the monitoring data to obtain a prevention and control ratio;

[0080] Performing causal analysis on the pest detection data according to gene drive to obtain an ecological risk degree, and constructing an optimized growth degree function according to the control ratio and the ecological risk degree;

[0081] An Internet of Things soybean cyst nematode population dynamic monitoring model is constructed according to the optimized growth degree function, the data to be monitored is input into the Internet of Things soybean cyst nematode population dynamic monitoring model, and the monitoring results are output.

[0082] In this embodiment, the method of performing target screening on the pest detection data to obtain target detection data includes:

[0083] The pest growth stage and the degree of crop damage and infection are obtained based on pest detection data. The pest development status is divided according to the number of pests at different growth stages. The pest danger level is obtained based on the degree of crop damage and infection. Pest data at key growth stages with high infection levels are prioritized, and pest individuals that are only in the egg stage or have a damage and infection level below 0.132 are eliminated to obtain primary data.

[0084] The gene drive matching degree is calculated based on the homology between the primary data and the target gene based on the pest gene sequence, and the primary data with a gene drive matching degree greater than the matching degree threshold is used as the key data;

[0085] Extract key features of key data, calculate the importance of key features using the random forest algorithm, sort key features in descending order based on importance, dynamically adjust the screening strategy based on drug resistance, and prioritize the key features with the highest scores;

[0086] Spatial autocorrelation analysis is performed on key data to identify pest clusters and obtain pest distribution hotspots. Spatiotemporal scanning statistical methods are used to detect specific time periods with rapid pest changes. Dissimilarity detection is performed on key data within specific time periods, and key data with a dissimilarity greater than 0.751 is output as target detection data.

[0087] In the actual evaluation, it was determined that the soybean cyst nematode was in the larval stage, the fruit damage infection level was 0.4, and the gene drive matching degree was calculated to be 0.85, and the key data were obtained; using the random forest algorithm for analysis, the characteristics of reproduction rate and diffusion speed were relatively important; spatial autocorrelation analysis found that the edge of the orchard was the gathering area of ​​soybean cyst nematodes; the spatiotemporal scanning statistical method determined that May to June was the period when the number of soybean cyst nematodes changed sharply, and the key data with a dissimilarity higher than 0.751 were used as target detection data.

[0088] In this embodiment, the method for performing a prevention and control risk assessment on the target detection data based on the monitoring data to obtain a prevention and control ratio includes:

[0089] Extract features from monitoring data to obtain monitoring features, select monitoring features based on spatiotemporal features to obtain spatiotemporal features;

[0090] Classification features are obtained by classifying target detection data, and interaction features are generated by matching spatiotemporal features with classification features. Classification features include pest characteristics, environmental characteristics, and gene drive association features. Pest characteristics include reproduction rate, spread speed, pesticide resistance index, and growth stage weight. Environmental characteristics include temperature and humidity adaptability and host plant coverage. Gene drive association features include target gene editing efficiency and gene drive spread potential.

[0091] Use ensemble learning methods to combine random forest and quarterly gradient boosting trees to build a prevention and control risk assessment model;

[0092] Calculate the degree of harm, controllability and ecological sensitivity based on the interaction characteristics:

[0093]

[0094]

[0095]

[0096] The hazard level is , the controllability is , the ecological sensitivity is , the weight coefficients are 、 、 、 、 , the reproduction rate is , the diffusion rate is The infection level is , the gene drive matching degree is , the target gene editing efficiency is , the resistance index is , the smoothing factor is , the host plant coverage is , the temperature and humidity difference is ;

[0097] Calculate the control ratio:

[0098]

[0099] The prevention ratio is , the risk weight is , the sensitive weight is ;

[0100] Optimize the learning rate of the prevention and control risk assessment model through grid search, input the target detection data into the optimized prevention and control risk assessment model, and output the prevention and control ratio;

[0101] In the actual assessment, pest characteristics were: reproduction rate of 0.9, spread speed of 0.7, resistance index of 0.3, and growth stage weight of 0.8; environmental characteristics: temperature and humidity adaptability of 0.8, host plant coverage of 0.7; gene drive association characteristics: target gene editing efficiency of 0.8, gene drive spread potential of 0.9;

[0102] Weight coefficient 、 、 、 、 They are 0.29, 0.21, 0.19, 0.16 and 0.15 respectively; the gene drive matching degree is 0.849, the target gene editing efficiency is 0.8, the resistance index is 0.3, the dynamic adjustment parameter is 0.2, the host plant coverage is 0.7, the temperature and humidity difference is 0.6, the hazard degree is 0.482, the ecological sensitivity is 0.138, the hazard weight is 0.172, the sensitive weight is 0.3, and the prevention and control ratio is 0.33.

[0103] In this embodiment, the method for performing causal analysis on the pest detection data based on gene drive to obtain ecological risk includes:

[0104] Obtain ecosystem data based on monitoring data, including abundance, population structure, food web interactions, and environmental parameters of non-target species;

[0105] Intervention variables are obtained based on the actual release rate of gene drive and the target pest population, and outcome variables are obtained based on the change rate of the target pest population and the changes in the abundance of non-target associated species; confounding variables are obtained based on environmental data and current control measures.

[0106] Confounding variables and mediating variables are added to refine the causal relationship. A causal graph based on a Bayesian network is constructed based on the refined causal relationship. When randomized controlled experiments are not possible, instrumental variables are introduced to perform anchored causal analysis to obtain enhanced causal relationships. A graph neural network is introduced to process the complex ecological network relationship based on the enhanced causal relationship to obtain the ecological causal relationship. The ecological network relationship is the causal relationship between pests, gene drives, and ecological functions. The mediating variable is the food web interaction relationship. The expression of the causal graph is:

[0107]

[0108] The outcome variable is Ot, the treatment variable is Tr, the confounding variable is Cf, and the marginal probability of the outcome variable is , the marginal probability of the treatment variable is , the conditional probability of the treatment variable under the confounding variable Cf is , the conditional probability distribution under the conditions of treatment variable Tr and confounding variable Cf is ;

[0109] Use the Bayesian formula to update the posterior distribution, the expression is:

[0110]

[0111] The Bayesian network parameters are , the new pest detection data is , the historical pest detection data is , Bayesian network parameters Download new pest detection data The probability of generating , new pest detection data Conditional Bayesian network parameters The posterior distribution of , historical pest detection data Conditional Bayesian network parameters The posterior distribution of ;

[0112] Calculate ecological risk based on objective weighting:

[0113]

[0114] The ecological risk is , the target pest reduction rate is The expected decline rate is , the non-target species impact index is , the ecosystem health index is , the ecological health weight is , the weight of non-target species is , the decline rate weight is ;

[0115] Through machine learning, the ecological health weight, non-target species weight, and decline rate weight are dynamically adjusted. The pest detection data is input into the adjusted causal diagram to output the ecological risk level.

[0116] In the actual assessment, the capture volume of gene-driven individuals in June 2023 was 500 per mu, the population change rate after release in 2022 was -35%, the target pest decline rate was 0.3, the expected decline rate was 0.5, the non-target species impact index was 0.2, the ecosystem health index was 0.8, the ecological health weight was 0.4, the non-target species weight was 0.3, the decline rate weight was 0.3, and the ecological risk was 0.56.

[0117] In this embodiment, the method for constructing an optimized growth degree function based on the control ratio and the ecological risk degree includes:

[0118] A directed weight mechanism is introduced to construct the control directed weight and ecological directed weight according to the control ratio and ecological risk. The expression is:

[0119]

[0120]

[0121] The prevention and control directed weight is , the ecological directed weight is , the prevention and treatment ratio at time t is The historical average prevention and control ratio is The standard deviation of the control ratio is , the ecological risk at time t is The safety threshold of ecological risk is The standard deviation of ecological risk is , the basic weight of the control ratio is , the basic weight of ecological risk is , the sensitivity coefficient is , the hyperbolic tangent function is ;

[0122] when When it is greater than zero, the control is effective; when When it is less than zero, the prevention and treatment is ineffective and the condition worsens; when When it is greater than zero, the ecological risk exceeds the safety range; when When it is less than , the ecological risk is controllable;

[0123] The optimal growth degree function is constructed based on the prevention and control directed weights and the ecological directed weights, and the expression is:

[0124]

[0125] The optimized growth function at time t is , the normalized prevention and control ratio at time t is , the normalized ecological risk at time t is ;

[0126] In the actual assessment, the prevention and control ratio was normalized to 0.6, the ecological risk was normalized to 0.7, the prevention and control directional weight was 0.29, the ecological directional weight was 0.88, and the degree of optimized development was 0.27.

[0127] In this embodiment, a method for constructing an IoT soybean cyst nematode population dynamic monitoring model based on the optimized growth degree function includes:

[0128] The objective weighted sum of the optimized growth degree function and the loss function is used as the objective function of the Internet of Things soybean cyst nematode population dynamics monitoring model.

[0129] The IoT-based soybean cyst nematode population dynamics monitoring model includes autoencoders, time series analysis algorithms, graph neural network algorithms, and multi-agent deep reinforcement learning;

[0130] The autoencoder encodes and compresses the input data into a low-dimensional representation, then decodes and reconstructs it. It automatically learns the intrinsic characteristics of the input data based on encoding and decoding to obtain the features to be analyzed.

[0131] Time series analysis algorithms build mathematical models to analyze the data variation patterns of the features to be analyzed at consecutive time points and capture time series characteristics. Time series characteristics include trends, seasonality, and periodicity.

[0132] The graph neural network algorithm converts time series characteristics into graph structure data, performs message passing and feature learning on the graph based on the objective function, captures the dynamic relationship of technological development, and realizes dynamic monitoring of priority development technologies;

[0133] Multi-agent deep reinforcement learning optimizes the Internet of Things soybean cyst nematode population dynamic monitoring model by having multiple agents interact and learn in the environment, and adjusting the strategy based on the feedback of the dynamic monitoring model.

[0134] Secondly, a dynamic monitoring system for priority development technologies based on intelligent algorithms, including:

[0135] Data acquisition module: used to collect pest detection data and monitoring data within the preset area of ​​gene-driven pest control, and pre-process the pest detection data and monitoring data; the monitoring data includes historical data and data to be analyzed; the data to be analyzed includes pest data, environmental data, and ecosystem data;

[0136] A screening risk assessment module is configured to perform target screening on the pest detection data to obtain target detection data, and perform control risk assessment on the target detection data based on the monitoring data to obtain a control ratio;

[0137] Causal fusion module: used to perform causal analysis on the pest detection data based on gene drive to obtain ecological risk, and construct an optimized growth degree function based on the control ratio and the ecological risk;

[0138] Modeling and optimization module: constructing an Internet of Things soybean cyst nematode population dynamic monitoring model according to the optimized growth degree function, inputting the data to be monitored into the Internet of Things soybean cyst nematode population dynamic monitoring model, and outputting the monitoring results.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for monitoring soybean cyst nematode population dynamics based on the Internet of Things, characterized in that: The following steps are involved: Collecting pest detection data and monitoring data within a preset area for gene-driven pest control, and pre-processing the pest detection data and monitoring data; the monitoring data includes historical data and data to be analyzed; the data to be analyzed includes pest data, environmental data, and ecosystem data; the pest detection data includes species, quantity, distribution location, infection level, growth stage of the pest, and gene sequence; the historical data includes control measures and historical monitoring data; Performing target screening on the pest detection data to obtain target detection data, and performing prevention and control risk assessment on the target detection data based on the monitoring data to obtain a prevention and control ratio; Performing causal analysis on the pest detection data according to gene drive to obtain an ecological risk degree, and constructing an optimized growth degree function according to the control ratio and the ecological risk degree; Constructing an IoT soybean cyst nematode population dynamic monitoring model according to the optimized growth degree function, inputting the data to be monitored into the IoT soybean cyst nematode population dynamic monitoring model, and outputting the monitoring results; The method for constructing an optimized growth degree function according to the control ratio and the ecological risk degree comprises: A directed weight mechanism is introduced to construct the control directed weight and ecological directed weight according to the control ratio and ecological risk. The expression is: The prevention and control directed weight is , the ecological directed weight is , the prevention and treatment ratio at time t is The historical average prevention and control ratio is The standard deviation of the control ratio is , the ecological risk at time t is The safety threshold of ecological risk is The standard deviation of ecological risk is , the basic weight of the control ratio is , the basic weight of ecological risk is , the sensitivity coefficient is , the hyperbolic tangent function is ; when When it is greater than zero, the control is effective; when When it is less than zero, the prevention and treatment is ineffective and the condition worsens; when When it is greater than zero, the ecological risk exceeds the safety range; when When it is less than , the ecological risk is controllable; The optimal growth degree function is constructed based on the prevention and control directed weights and the ecological directed weights, and the expression is: The optimized growth function at time t is , the normalized prevention and control ratio at time t is , the normalized ecological risk at time t is .

2. The method for monitoring soybean cyst nematode population dynamics based on the Internet of Things according to claim 1, characterized in that: The method of performing target screening on the pest detection data to obtain target detection data comprises: The pest growth stage and the degree of crop damage and infection are obtained based on pest detection data. The pest development status is divided according to the number of pests at different growth stages. The pest danger level is obtained based on the degree of crop damage and infection. Pest data at key growth stages with high infection levels are prioritized, and pest individuals that are only in the egg stage or have a damage and infection level below 0.132 are eliminated to obtain primary data. The gene drive matching degree is calculated based on the homology between the primary data and the target gene based on the pest gene sequence, and the primary data with a gene drive matching degree greater than the matching degree threshold is used as the key data; Extract key features of key data, calculate the importance of key features using the random forest algorithm, sort key features in descending order based on importance, dynamically adjust the screening strategy based on drug resistance, and prioritize the key features with the highest scores; Spatial autocorrelation analysis was performed on key data to identify pest clustering areas and obtain pest distribution hotspots. Spatiotemporal scanning statistical methods were used to detect specific time periods with drastic changes in pests. Dissimilarity detection was performed on key data within specific time periods, and key data with a dissimilarity higher than 0.751 were output as target detection data.

3. The method for monitoring soybean cyst nematode population dynamics based on the Internet of Things according to claim 1, characterized in that: The method for performing a prevention and control risk assessment on the target detection data based on the monitoring data to obtain a prevention and control ratio includes: Extract features from monitoring data to obtain monitoring features, select monitoring features based on spatiotemporal features to obtain spatiotemporal features; Classification features are obtained by classifying target detection data, and interaction features are generated by matching spatiotemporal features with classification features. Classification features include pest characteristics, environmental characteristics, and gene drive association features. Pest characteristics include reproduction rate, spread speed, pesticide resistance index, and growth stage weight. Environmental characteristics include temperature and humidity adaptability and host plant coverage. Gene drive association features include target gene editing efficiency and gene drive spread potential. Use ensemble learning methods to combine random forest and quarterly gradient boosting trees to build a prevention and control risk assessment model; Calculate the degree of harm, controllability and ecological sensitivity based on the interaction characteristics: The hazard level is , the controllability is , the ecological sensitivity is , the weight coefficients are 、 、 、 、 , the reproduction rate is , the diffusion rate is The infection level is , the gene drive matching degree is , the target gene editing efficiency is , the resistance index is , the smoothing factor is , the host plant coverage is , the temperature and humidity difference is ; Calculate the control ratio: The prevention ratio is , the risk weight is , the sensitive weight is ; The learning rate of the prevention and control risk assessment model is optimized through grid search, the target detection data is input into the optimized prevention and control risk assessment model, and the prevention and control ratio is output.

4. The method for monitoring soybean cyst nematode population dynamics based on the Internet of Things according to claim 1, characterized in that: The method for performing causal analysis on the pest detection data based on gene drive to obtain an ecological risk level includes: Obtain ecosystem data based on monitoring data, including abundance, population structure, food web interactions, and environmental parameters of non-target species; Intervention variables are obtained based on the actual release rate of gene drive and the target pest population, and outcome variables are obtained based on the change rate of the target pest population and the changes in the abundance of non-target associated species; confounding variables are obtained based on environmental data and current control measures. Confounding variables and mediating variables are added to refine the causal relationship. A causal graph based on a Bayesian network is constructed based on the refined causal relationship. When randomized controlled experiments are not possible, instrumental variables are introduced to perform anchored causal analysis to obtain enhanced causal relationships. A graph neural network is introduced to process the complex ecological network relationship based on the enhanced causal relationship to obtain the ecological causal relationship. The ecological network relationship is the causal relationship between pests, gene drives, and ecological functions. The mediating variable is the food web interaction relationship. The expression of the causal graph is: The outcome variable is Ot, the treatment variable is Tr, the confounding variable is Cf, and the marginal probability of the outcome variable is , the marginal probability of the treatment variable is , the conditional probability of the treatment variable under the confounding variable Cf is , the conditional probability distribution under the conditions of treatment variable Tr and confounding variable Cf is ; Use the Bayesian formula to update the posterior distribution, the expression is: The Bayesian network parameters are , the new pest detection data is , the historical pest detection data is , Bayesian network parameters Download new pest detection data The probability of generating , new pest detection data Conditional Bayesian network parameters The posterior distribution of , historical pest detection data Conditional Bayesian network parameters The posterior distribution of ; Calculate ecological risk based on objective weighting: The ecological risk is , the target pest reduction rate is The expected decline rate is , the non-target species impact index is , the ecosystem health index is , the ecological health weight is , the weight of non-target species is , the decline rate weight is ; Through machine learning, the ecological health weight, non-target species weight, and decline rate weight are dynamically adjusted, and the pest detection data is input into the adjusted causal diagram to output the ecological risk level.

5. The method for monitoring soybean cyst nematode population dynamics based on the Internet of Things according to claim 1, characterized in that: The method for constructing an Internet of Things soybean cyst nematode population dynamic monitoring model based on the optimized growth degree function includes: The objective weighted sum of the optimized growth degree function and the loss function is used as the objective function of the Internet of Things soybean cyst nematode population dynamics monitoring model. The IoT-based soybean cyst nematode population dynamics monitoring model includes autoencoders, time series analysis algorithms, graph neural network algorithms, and multi-agent deep reinforcement learning; The autoencoder encodes and compresses the input data into a low-dimensional representation, then decodes and reconstructs it. It automatically learns the intrinsic characteristics of the input data based on encoding and decoding to obtain the features to be analyzed. Time series analysis algorithms build mathematical models to analyze the data variation patterns of the features to be analyzed at consecutive time points and capture time series characteristics. Time series characteristics include trends, seasonality, and periodicity. The graph neural network algorithm converts time series characteristics into graph structure data, performs message passing and feature learning on the graph based on the objective function, captures the dynamic relationship of technological development, and realizes dynamic monitoring of priority development technologies; Multi-agent deep reinforcement learning optimizes the Internet of Things soybean cyst nematode population dynamic monitoring model by having multiple agents interact and learn in the environment, and adjusting the strategy based on the feedback of the dynamic monitoring model.

6. A soybean cyst nematode population dynamics monitoring system based on the Internet of Things, used to implement the method according to any one of claims 1 to 5, characterized in that: include: Data acquisition module: used to collect pest detection data and monitoring data within the preset area of ​​gene-driven pest control, and pre-process the pest detection data and monitoring data; the monitoring data includes historical data and data to be analyzed; the data to be analyzed includes pest data, environmental data, and ecosystem data; A screening risk assessment module is configured to perform target screening on the pest detection data to obtain target detection data, and perform control risk assessment on the target detection data based on the monitoring data to obtain a control ratio; Causal fusion module: used to perform causal analysis on the pest detection data based on gene drive to obtain ecological risk, and construct an optimized growth degree function based on the control ratio and the ecological risk; Modeling and optimization module: constructing an Internet of Things soybean cyst nematode population dynamic monitoring model according to the optimized growth degree function, inputting the data to be monitored into the Internet of Things soybean cyst nematode population dynamic monitoring model, and outputting the monitoring results.

Images