Pesticide risk prevention and control system for rice and shrimp co-cropping in cold region
The cold-region rice-shrimp co-cultivation pesticide risk control system, which combines multi-source sensors and high-resolution imaging equipment with deep learning models and knowledge graphs, solves the problems of accuracy and timeliness in the prevention and control of diseases, pests and weeds in the cold-region rice-shrimp co-cultivation model. It achieves precise identification and scientific application of pesticides, reduces pesticide risks, and protects the ecological environment of paddy fields.
Patent Information
- Application Number
- CN202511243932.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-02
- Publication Date
- 2025-11-21
AI Technical Summary
In the cold-region rice-crayfish co-cultivation model, the accuracy and timeliness of pest and weed control are low, and the use of pesticides is unscientific, which affects the quality of rice and the living environment of crayfish. Moreover, the existing equipment is difficult to adapt to the cold climate conditions.
Data is collected using a multi-source sensor array and high-resolution imaging equipment. Combined with deep learning models and knowledge graphs, the intelligent analysis module identifies pests, diseases, and weeds, the decision-making module formulates control strategies, and the automated equipment implements precise pesticide application.
It enables accurate identification and prediction of pests, diseases, and weeds, allows for the scientific formulation of pesticide application plans, reduces pesticide use, ensures the safe production of rice and crayfish, and maintains the ecological balance of paddy fields.
Smart Images
Figure CN120996584A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of agricultural pest control, specifically to a cold region rice and crayfish co-cultivation pesticide risk prevention and control system. BACKGROUND
[0002] Cold region rice and crayfish co-cultivation is an ecological agricultural model that combines rice cultivation and crayfish breeding, with both economic and ecological benefits. However, in this model, pest control faces many challenges.
[0003] Traditional pest control methods rely mainly on farmers' experience and judgment, which has the following problems: first, the accuracy is low, farmers are difficult to accurately identify the type and stage of pest occurrence, which can easily lead to misjudgment; second, the timeliness is poor, it is difficult to monitor the occurrence of pests in real time, often missing the best control period; third, the use of pesticides is not scientific, in order to quickly control pests, farmers may overuse pesticides, which not only affects the quality of rice and the survival environment of crayfish, but also destroys the ecological balance of the rice field, causing serious pesticide risks.
[0004] In addition, the climate conditions in cold regions are special, with a long period of low temperature, and the occurrence patterns of pests differ from other regions. The existing control devices and methods are difficult to meet the specific needs of cold region rice and crayfish co-cultivation. Therefore, there is an urgent need for a pesticide risk prevention and control device that can accurately predict and identify rice field pests and develop a scientific control plan. SUMMARY
[0005] The purpose of the present application is to provide a cold region rice and crayfish co-cultivation pesticide risk prevention and control system to solve the problems of low accuracy, poor timeliness, and unscientific use of pesticides in the existing cold region rice and crayfish co-cultivation model, and to achieve accurate prediction, identification, and scientific control of rice field pests, reduce pesticide risks, and ensure the safe production of rice and crayfish.
[0006] The purpose of the present application can be achieved by the following technical solutions:
[0007] A cold region rice and crayfish co-cultivation pesticide risk prevention and control system, comprising the following steps:
[0008] Sensing module: composed of a multi-source sensor array and a high-resolution imaging device, used to collect meteorological parameters, hydrological parameters, soil characteristics, and visual information of pests and diseases in the rice field;
[0009] Intelligent analysis module: connected to the sensing module, receives and processes data from the sensing module, classifies and predicts pests and diseases through a deep learning model, and evaluates their occurrence trends and potential hazards based on environmental parameters;
[0010] Optimization decision module: connected with intelligent analysis module, built-in disease and pest control knowledge graph in cold rice-shrimp co-culture mode, formulates optimal prevention and control strategy according to analysis results, and comprehensively considers environmental impact, economic cost and prevention and control effect;
[0011] Execution control module: connected with optimization decision module, including automatic equipment and remote communication interface, used for implementing prevention and control measures or sending operation instructions to farmers.
[0012] As a further scheme of the application: the multi-source sensor array has a low-temperature compensation function and can work stably in the range of-30 DEG C to 40 DEG C, ensuring the reliability of data acquisition in cold environments.
[0013] As a further scheme of the application: the high-resolution imaging device supports wide-angle shooting and macro shooting, can capture the overall condition and local details of the rice field, and has infrared imaging capability for disease and pest detection at night or in low light conditions.
[0014] As a further scheme of the application: the deep learning model is based on an improved YOLOv5 architecture and is customized for cold rice-shrimp co-culture scenarios, and can identify common diseases and pests such as rice planthoppers, negative mud bugs, and rice blast, as well as weed types such as sedge and barnyard grass.
[0015] As a further scheme of the application: the knowledge graph contains life cycle information, environmental sensitivity characteristics and prevention and control measure association rules of diseases and pests, and supports dynamic updating and expansion.
[0016] As a further scheme of the application: the automatic equipment is equipped with an adaptive spraying system that adjusts the spraying amount and angle according to the distribution density of diseases and pests, reducing pesticide waste and environmental pollution.
[0017] As a further scheme of the application: the intelligent analysis module includes:
[0018] Data preprocessing:
[0019] The environmental parameters collected by the sensor are standardized, and the Z-score standardization method is used, with the formula: x'=(x-μ) / σ;
[0020] Wherein, x is the original data, μ is the mean, and σ is the standard deviation;
[0021] The image captured by the camera is preprocessed, including image denoising, image enhancement and image segmentation;
[0022] Feature extraction:
[0023] The improved ResNet-50 convolutional neural network is used for feature extraction of the preprocessed image, and the attention mechanism is introduced to enhance the weight of key features;
[0024] The time sequence feature extraction method based on LSTM is adopted for the environmental parameters to capture the change rule of the environmental parameters over time;
[0025] Identification and classification:
[0026] The extracted image features and environmental parameter features are fused, and a fusion feature vector is obtained by using feature splicing;
[0027] The fusion feature vector is classified by using a support vector machine classifier to identify the types of diseases, pests and weeds;
[0028] A multi-label classification algorithm is used to determine the occurrence degree and stage of diseases, pests and weeds.
[0029] As a further scheme of the application, the intelligent analysis module further comprises:
[0030] The training set and the test set are divided by using cross-validation, and the proportion is 7:3;
[0031] The Adam optimizer is used for model training, the initial learning rate is set to 0.001, and the learning rate is adjusted by using a cosine annealing strategy as the training rounds increase;
[0032] The model performance is evaluated by calculating the precision, recall and F1 score, and the training is stopped when the F1 score does not improve for 5 consecutive rounds.
[0033] As a further scheme of the application, the execution control module specifically comprises:
[0034] Rule reasoning:
[0035] The decision rule base is constructed based on the production rule, and the rule form is: I F (condition 1) AND (condition 2)... THEN (conclusion);
[0036] The identification result of the AI processing unit is matched with the conditions in the rule base by using a forward reasoning mechanism to obtain a preliminary prevention and control scheme;
[0037] Reinforcement learning optimization:
[0038] A Markov decision process model is constructed, the state space includes the types, occurrence degree, stage and current environmental parameters of diseases, pests and weeds, the action space is various possible prevention and control measures, and the reward function is defined as:
[0039] R = alpha x (1-pesticide usage amount / maximum allowable usage amount) + beta x (disease and pest control effect) + gamma x (crayfish survival environment influence coefficient);
[0040] Wherein, alpha, beta and gamma are weight coefficients, and alpha + beta + gamma = 1;
[0041] The preliminary prevention and treatment scheme is optimized by using a deep reinforcement learning algorithm, and the Q value network is continuously updated through interaction with the environment to obtain an optimal pesticide application scheme.
[0042] As a further scheme of the application, the execution control module further comprises:
[0043] The weights of each target are determined by using an analytic hierarchy process method by comprehensively considering the prevention and treatment effect, pesticide usage and influence on the survival environment of crayfish.
[0044] The multiple alternative schemes are ranked by using a TOPSIS method, and the optimal pesticide application scheme is selected.
[0045] The beneficial effects of the application are as follows:
[0046] The sensor of the monitoring unit of the application collects the environmental parameters of the rice field in real time, the camera regularly shoots the image of the rice field, and the collected data and image information are transmitted to the AI processing unit.
[0047] After receiving the data and image information, the AI processing unit pre-processes, extracts features, identifies and classifies the data and image according to the above algorithm, determines the type, occurrence degree and stage of the disease, pest and weed, and transmits the result to the decision unit.
[0048] According to the result of the AI processing unit, the decision unit first obtains a preliminary prevention and treatment scheme through rule-based reasoning, then optimizes it by using reinforcement learning, and finally selects the optimal pesticide application scheme through multi-objective decision.
[0049] The communication module of the execution unit sends the pesticide application scheme that needs manual operation (such as releasing trichogramma) to the farmer, and the pesticide application equipment performs precise pesticide application according to the automatic pesticide application scheme.
[0050] The application realizes accurate identification and prediction of diseases, pests and weeds: through the deep learning model and specific algorithm of the AI processing unit, combined with the environmental parameters and image information collected by the monitoring unit, the type, occurrence degree and stage of the disease, pest and weed can be accurately identified, improving the accuracy and timeliness of identification.
[0051] Scientifically formulate the pesticide application scheme: the decision unit formulates a targeted pesticide application scheme according to the identification result and the built-in prevention and treatment database through specific rule-based reasoning, reinforcement learning optimization and multi-objective decision algorithm, avoiding blind use of pesticides, reducing pesticide usage and reducing pesticide risk.
[0052] Adapt to the cold rice and crayfish co-cultivation environment: the sensors and cameras of the device are designed to be waterproof and cold-resistant, and can work stably under the special climate conditions of cold land, meeting the needs of the cold rice and crayfish co-cultivation mode.
[0053] Improve the efficiency of prevention and treatment: the application equipment of the execution unit can carry out precise application, and the communication module timely delivers manual operation instructions, thereby improving the efficiency of prevention and treatment of diseases, pests and weeds and guaranteeing the safe production of rice and crayfish.
[0054] Protect the ecological environment: reduce the use of pesticides, reduce the damage to the ecological environment of the rice field, and be conducive to maintaining the ecological balance of the cold rice and crayfish system. BRIEF DESCRIPTION OF DRAWINGS
[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.
[0056] Figure 1 The present application provides a system block diagram of a cold rice and crayfish co-cultivation pesticide risk prevention and control system. DETAILED DESCRIPTION
[0057] In order to make the person skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.
[0058] As shown in Figure 1 The cold rice and crayfish co-cultivation pesticide risk prevention and control system provided by the present application specifically comprises the following steps:
[0059] The perception module is composed of a multi-source sensor array and a high-resolution imaging device, which is used to collect meteorological parameters (such as temperature, humidity, wind speed), hydrological parameters (such as water level, dissolved oxygen concentration), soil properties (such as pH, conductivity) and visual information of diseases, pests and weeds in the rice field.
[0060] In the perception module:
[0061] The multi-source sensor array has a low-temperature compensation function and can work stably in the range of-30℃ to 40℃, ensuring the reliability of data collection in cold environments.
[0062] The high-resolution imaging device supports wide-angle shooting and macro shooting, can capture the overall condition and local details of the rice field, and has infrared imaging capability for detection of diseases, pests and weeds in night or low light conditions.
[0063] Intelligent analysis module: connected with the perception module, receiving and processing the data transmitted by the perception module, classifying and predicting the diseases and pests through the deep learning model, and evaluating the occurrence trend and potential harm in combination with the environmental parameters.
[0064] In the intelligent analysis module:
[0065] The deep learning model is based on the improved YOLOv5 architecture, trained by a large number of images of diseases and pests in cold rice-shrimp co-culture rice fields and corresponding environmental parameters, and can accurately identify common diseases and pests such as Chilo suppressalis, rice planthopper, rice leaf roller, rice water weevil, barnyard grass, leafflower, sedge, broadleaf weeds, rice blast, sheath blight, and rice false smut.
[0066] The specific process of classifying and predicting diseases and pests is as follows:
[0067] Data preprocessing:
[0068] The environmental parameters collected by the sensor are standardized, using the Z-score standardization method, the formula is: x' = (x-μ) / σ.
[0069] Where x is the original data, μ is the mean, and σ is the standard deviation.
[0070] The image taken by the camera is preprocessed, including image denoising (using median filter algorithm), image enhancement (using histogram equalization algorithm) and image segmentation (using Otsu algorithm based on threshold).
[0071] Feature extraction:
[0072] The improved ResNet-50 convolutional neural network is used to extract features from the preprocessed image, and the attention mechanism (SE module) is introduced to enhance the weight of key features.
[0073] The time series feature extraction method based on LSTM is used for environmental parameters to capture the change rule of environmental parameters over time.
[0074] Identification and classification:
[0075] The extracted image features and environmental parameter features are fused, and the fusion feature vector is obtained by feature splicing.
[0076] The support vector machine (SVM) classifier is used to classify the fusion feature vector to identify the type of diseases and pests.
[0077] The multi-label classification algorithm (ML-KNN) is used to judge the occurrence degree and stage of diseases and pests.
[0078] Further, a time series prediction model (LSTM) is used to predict the occurrence probability of plant diseases, insect pests and weeds in the next 7 days, and the formula is:
[0079] P(t+1) = σ(W h h t +W x x t +b)
[0080] Where P(t+1) represents the occurrence probability at the next time, h t is the hidden state, x t is the input feature, W h and W x are weight matrices, b is the bias term, and σ is the activation function.
[0081] It also includes model training and optimization:
[0082] The data set is divided by K-fold cross-validation, and K is set to 5.
[0083] The RMSprop optimizer is used for model training, and the initial learning rate is set to 0.0005. After every 10 training rounds, the learning rate is adjusted according to the exponential decay strategy, and the formula is:
[0084] η t =η0*e -λt
[0085] Where η t is the current learning rate, η0 is the initial learning rate, λ is the decay coefficient, and t is the training round.
[0086] Optimization decision module: connected with the intelligent analysis module, built-in disease, insect pest and weed control knowledge graph in cold rice-shrimp co-culture mode, according to the analysis results to develop the optimal control strategy, comprehensive consideration of environmental impact, economic cost and control effect.
[0087] In the optimization decision module:
[0088] The knowledge graph contains the life cycle information of plant diseases, insect pests and weeds, environmental sensitivity characteristics and prevention and control measures associated rules, and supports dynamic updating and expansion.
[0089] Wherein the process of reinforcement learning optimization is:
[0090] A Markov decision process (MDP) model is constructed, the state space includes the types, occurrence degree, current stage and environmental parameters of plant diseases, insect pests and weeds, the action space is various possible control measures, and the reward function is defined as:
[0091] R = α × (1-pesticide use amount / maximum allowable use amount) + β × (disease and pest control effect) + γ × (crayfish survival environment influence coefficient).
[0092] wherein a, b, g are weight coefficients, and a+b+g = 1.
[0093] A deep reinforcement learning algorithm (DQN) is used to optimize the preliminary control scheme, and the Q-value network is continuously updated through interaction with the environment to obtain the optimal pesticide application scheme.
[0094] Multi-objective decision making:
[0095] Considering the prevention and control effect, pesticide use and the impact on the survival environment of crayfish, the analytic hierarchy process (AHP) is used to determine the weight of each target.
[0096] TOPSIS method is used to sort multiple alternative schemes and select the optimal pesticide application scheme.
[0097] Execution control module: connected with the optimization decision module, including automated equipment (such as drones, spraying robots) and remote communication interface, used to implement control measures or send operation instructions to farmers.
[0098] The running process of the system is described as follows in combination with specific application scenarios:
[0099] When the AI processing unit identifies that the rice stem borer is in the oviposition period, the rule reasoning module of the decision unit matches the corresponding rule to obtain the preliminary scheme of releasing Trichogramma. The reinforcement learning module takes the current rice stem borer oviposition situation and environmental parameters as the state, and the number and frequency of releasing Trichogramma as the action, and optimizes to obtain a scheme of releasing 150,000 Trichogramma per hectare in three times. The communication module of the execution unit sends the instruction to the farmer's mobile phone APP, and the farmer releases Trichogramma according to the instruction.
[0100] When the AI processing unit identifies that the rice stem borer is in the low-age larva period, the rule reasoning module of the decision unit obtains the preliminary scheme of applying Bacillus thuringiensis (Bt) preparation. The reinforcement learning module considers the current larva quantity and environmental temperature parameters, optimizes the pesticide dosage and concentration, and finally formulates a suitable pesticide application scheme. The pesticide application equipment of the execution unit automatically sprays Bacillus thuringiensis (Bt) preparation according to the scheme.
[0101] When the AI processing unit identifies that the rice planthopper exists and the number of hundred plants reaches 1000, the rule reasoning module of the decision unit matches the rules of applying pymetrozine and thiamethoxam. The multi-objective decision making module considers the prevention and control effect and the impact on crayfish to determine the appropriate pesticide ratio and application time, and the pesticide application equipment of the execution unit applies pesticides according to the corresponding scheme. If the number of hundred plants does not reach 1000, the rule reasoning module obtains the preliminary scheme of applying Beauveria bassiana and matrine, and after optimization by the reinforcement learning, the pesticide application equipment of the execution unit operates according to the optimized scheme.
[0102] For weed control, when the AI processing unit identifies barnyard grass and leafflower, the rule inference module of the decision unit obtains the initial plan of spraying with penoxsulam. The reinforcement learning module optimizes the moving path and dosage of the spraying equipment according to the density and growth stage of the weeds, and the spraying equipment of the execution unit operates according to the optimized plan; when it identifies sedge and broadleaf weeds, it similarly formulates the optimized plan of using bromofenoxim and halosulfuron, and the spraying equipment of the execution unit operates according to the plan.
[0103] For disease control, when the AI processing unit identifies that the rice blast is in the early stage of the disease (5% of the diseased leaves), or the panicle neck blast is 3-5 days before the break (the key prevention period) and the heading stage, the rule inference module of the decision unit calls the initial plan of using tricyclazole, anilofos and kasugamycin. The multi-objective decision module considers the spread speed of the disease and the influence of the pesticide on the rice field ecology to determine the best spraying interval and dosage, and the spraying equipment of the execution unit sprays.
[0104] The above describes one embodiment of the present application in detail, but the content described is only the preferred embodiment of the present application, and cannot be considered as limiting the scope of the present application. Any equivalent changes and improvements made according to the scope of the present application should still belong to the scope of the present application.
Claims
1. A cold rice and shrimp co-production pesticide risk prevention and control system, characterized in that, The method comprises the following steps: The perception module is composed of a multi-source sensor array and a high-resolution imaging device, which is used to collect meteorological parameters, hydrological parameters, soil characteristics and visual information of diseases, pests and weeds in the rice field; The intelligent analysis module is connected with the perception module, receives and processes the data transmitted by the perception module, classifies and predicts diseases, pests and weeds through a deep learning model, and evaluates their occurrence trend and potential harm in combination with environmental parameters; The optimization decision module is connected with the intelligent analysis module, and has a built-in knowledge graph of disease, pest and weed control in the cold rice-shrimp co-culture mode, and formulates the optimal control strategy according to the analysis result, and comprehensively considers the environmental impact, economic cost and control effect; The execution control module is connected with the optimization decision module, and comprises automatic equipment and a remote communication interface, which is used to implement the control measures or send operation instructions to farmers.
2. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The multi-source sensor array has a low-temperature compensation function, and can work stably in the range of-30℃ to 40℃, ensuring the reliability of data collection in the cold environment.
3. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The high-resolution imaging device supports wide-angle shooting and macro shooting, can capture the overall condition and local details of the rice field, and has infrared imaging capability for detecting diseases, pests and weeds in night or low light conditions.
4. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The deep learning model is based on the improved YOLOv5 architecture, and is customized for the cold rice-shrimp co-culture scene, and can identify common diseases, pests and weeds such as rice planthopper, negative mud bug and rice blast, as well as weed types such as sedge and barnyard grass.
5. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The knowledge graph contains the life cycle information, environmental sensitivity characteristics and control measure correlation rules of diseases, pests and weeds, and supports dynamic updating and extension.
6. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The automatic equipment is equipped with an adaptive spraying system, which adjusts the spraying amount and spraying angle according to the distribution density of diseases, pests and weeds, reducing pesticide waste and environmental pollution.
7. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The intelligent analysis module comprises: Data preprocessing: Standardize the environmental parameters collected by the sensor, and use the Z-score standardization method, the formula is: x'=(x-μ) / σ; Wherein, x is the original data, μ is the mean, and σ is the standard deviation; Preprocess the images taken by the camera, including image denoising, image enhancement and image segmentation; Feature extraction: An improved ResNet-50 convolutional neural network is used to extract features from the preprocessed images, and the attention mechanism is introduced to enhance the weight of key features; The time series feature extraction method based on LSTM is used for environmental parameters to capture the change rule of environmental parameters over time; Identification and classification: Fuse the extracted image features and environmental parameter features to obtain a fusion feature vector by feature splicing; Use a support vector machine classifier to classify the fusion feature vector to identify the types of diseases, pests and weeds; Use a multi-label classification algorithm to determine the occurrence degree and stage of diseases, pests and weeds.
8. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 7, characterized in that, The intelligent analysis module further comprises: The training set and test set are divided by cross-validation, with a ratio of 7:3; The Adam optimizer is used for model training, with an initial learning rate of 0.001, and the learning rate is adjusted using the cosine annealing strategy as the training rounds increase; Stop training when the F1 score does not improve for 5 consecutive rounds.
9. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 1, characterized in that, The execution control module comprises: Rule reasoning: The decision rule base is constructed based on the production rule, and the rule form is: I F (condition 1) AND (condition 2)... THEN (conclusion); The recognition result of the AI processing unit is matched with the conditions in the rule base by using the forward reasoning mechanism, and the preliminary prevention and control scheme is obtained; Reinforcement learning optimization: A Markov decision process model is constructed, the state space includes the types, occurrence degree, current stage and environmental parameters of the disease, pest and weed, the action space is various possible prevention and control measures, and the reward function is defined as: R = α × (1-pesticide use amount / maximum allowable amount) + β × (disease, pest and weed control effect) + γ × (crayfish survival environment influence coefficient); Wherein α, β, γ are weight coefficients, and α + β + γ = 1; The deep reinforcement learning algorithm is used to optimize the preliminary prevention and control scheme, and the Q value network is updated through interaction with the environment to obtain the optimal pesticide application scheme.
10. The cold region rice and shrimp co-culture pesticide risk prevention and control system according to claim 9, characterized in that, The execution control module further comprises: Considering the prevention and control effect, pesticide use amount and the influence on crayfish survival environment, the weight of each target is determined by using the analytic hierarchy process; TOPSIS method is used to sort multiple alternative schemes, and the optimal pesticide application scheme is selected.