5G-based pest killing system and method

Abstract

The invention relates to the technical field of intelligent agriculture, and discloses a 5G-based pest killing system and method.The 5G-based pest killing method comprises the steps that pest and environment data are collected through a miniature biological activity sensor network, a biological control agent formula is optimized by adopting a multi-objective evolutionary algorithm, a preference guiding mechanism is introduced for multi-objective tradeoff, and a pest killing result is obtained. An intelligent synthesis unit and precise putting equipment are remotely controlled through a 5G network, an optimal putting strategy is determined by operating an environment adaptability microorganism putting algorithm, and control effect data are continuously collected for closed-loop optimization; according to the invention, customized synthesis and precise release of the biological control agent are realized, the pest control efficiency is obviously improved, the resource consumption is reduced, the influence on the ecological environment is reduced, the system adaptability is enhanced, and the intelligent and precise development of pest control is comprehensively promoted.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture technology, and more specifically, to a 5G-based pest control system and method. Background Technology

[0002] With the intensification and large-scale development of agricultural production, pest control plays a vital role in ensuring the safety and stability of agricultural production. Traditional pest control methods mainly rely on the large-scale spraying of chemical pesticides. While achieving certain control effects, this has also brought a series of problems such as environmental pollution, ecological damage, and food safety risks. In recent years, biological control, as an environmentally friendly pest control method, has gradually gained attention and application.

[0003] Existing biological control technologies mainly include natural enemy biological control, microbial agent control, and plant-derived biopesticide control. While these technologies are more environmentally friendly than chemical pesticides, they still face a series of technical challenges: traditional biological control often uses general-purpose agents, which cannot be optimized for specific pest species or environmental conditions, resulting in unstable control effects and difficulty in dealing with new pests or resistance variations; existing systems overemphasize single objectives (such as killing efficiency) in the design of biological control agents, neglecting the balance between multiple objectives such as environmental friendliness, persistence, and cost, and failing to find the optimal formulation combination under complex constraints; they lack specificity and precision, and cannot make intelligent adjustments based on environmental conditions and pest distribution, leading to unstable control effects and low efficiency; large-scale uniform application lacks directionality, making it difficult to achieve precise application and real-time monitoring, resulting in significant resource waste; and they lack a complete closed-loop monitoring and adjustment mechanism, making it impossible to evaluate control effects in real time and adjust strategies promptly, especially when environmental conditions change rapidly or pest activity is highly variable, the control system cannot respond quickly and make corresponding adjustments.

[0004] The development of information technologies such as 5G, artificial intelligence, and edge computing has provided new technical means to solve the aforementioned problems. However, there is currently a lack of a complete solution that can fully leverage the advantages of 5G networks to achieve intelligent optimization, precise application, and real-time monitoring of biological control agent formulations. Therefore, how to utilize 5G technology and artificial intelligence algorithms to build an efficient, environmentally friendly, and intelligent pest control system has become an urgent technical problem to be solved. Summary of the Invention

[0005] This invention provides a 5G-based pest control system and method, which solves the technical problems in related technologies such as the inability to optimize biological control agent formulations, lack of precision in application, and inability to adjust in real time.

[0006] This invention provides a 5G-based pest control method, comprising the following steps: A network of miniature bioactive sensors is used to collect data on pest density, species, behavioral characteristics, and environmental parameters, and the data is transmitted to the processing system via a 5G network. Based on pest and environmental data received by the processing system, a multi-objective evolutionary algorithm is used to optimize the formulation of biological control agents. The control problem is formalized into a multi-objective optimization problem that includes control efficiency, environmental impact, persistence and economic cost. A preference guidance mechanism is introduced to weigh the multiple objectives. The optimized formula instructions are transmitted to the intelligent synthesis unit via 5G network to produce customized biological control agents on site; based on pest and environmental data in the processing system, an environmentally adaptive microbial delivery algorithm is run to generate a pest distribution heat map and calculate the best delivery strategy, and the produced biological control agents are released by the precision delivery equipment controlled by 5G network. The system continuously collects data on prevention and control effects through a network of miniature bioactive sensors, and feeds this data back to the central data center via a 5G network. Based on the feedback data, the formulation and application strategy are optimized.

[0007] In a preferred embodiment, the preference guidance mechanism dynamically adjusts the weight of a single target and the interaction weight between targets according to the application scenario; The environmentally adaptive microbial delivery algorithm considers three key dimensions: delivery location, time, and dosage, to calculate the globally optimal delivery strategy.

[0008] In a preferred embodiment, the micro bioactive sensor network includes optical sensors, acoustic sensors, chemical sensors, environmental sensors, and soil sensors, used to collect basic information about pests, behavioral characteristic data, resistance characteristics, and environmental correlation data; the collected data is preprocessed through edge computing nodes and then transmitted to a central data center for in-depth analysis.

[0009] In a preferred embodiment, the multi-objective evolutionary algorithm is a non-dominated sorting genetic algorithm, comprising: Formula parameters are represented using real number encoding; Individual selection is performed by combining non-dominated ranking and reference point methods; An adaptive mutation operation is introduced, and the mutation rate is dynamically adjusted according to the population diversity. Introducing a local search mechanism in the later stages of evolution improves the accuracy of the solution; Output a set of Pareto undominated solutions, each solution representing an optimal recipe under multiple objective trade-offs.

[0010] In a preferred embodiment, the preference guidance mechanism comprehensively evaluates multiple objectives through a comprehensive utility function and dynamically adjusts the weights in the following manner: For ecologically sensitive areas, increase the weight of environmental impact targets; The weighting of the goal of improving control efficiency during pest outbreak periods; For areas requiring long-term prevention and control, increase the weight of persistent objectives; For cash crops, increase the weighting of cost and efficiency objectives.

[0011] In a preferred embodiment, the environmentally adaptive microbial delivery algorithm includes: Construct a spatial data model of pest distribution and generate a heat map of pest distribution; Determine the coordinate set of release locations, prioritizing areas with high pest density and key diffusion nodes; Determine the set of release time points and select the best release time based on pest activity patterns and environmental factors; Calculate the optimal dosage for each location, and dynamically adjust the dosage based on pest density and environmental conditions; The global optimal deployment strategy is solved using the simulated annealing optimization method.

[0012] In a preferred embodiment, the smart synthesis unit performs the following steps to produce customized biocontrol agents: The optimized formula parameters are converted into specific executable instructions, including the precise amount of each component to be added, the order of addition, reaction time and temperature control parameters; Accurately measure each component according to instructions, and mix the components in the specified order and under the specified conditions; Add stabilizers and perform necessary physical treatments to ensure the stability of the control agent; By using built-in sensors to monitor key parameters of the synthesis process in real time, we ensure that product quality meets expectations. The synthesized control agents are evaluated from multiple dimensions, including physical properties, chemical properties, and biological activity, and the test results are fed back to the system center.

[0013] In a preferred embodiment, the continuous collection of prevention and control effect data includes: Construct a multi-indicator prevention and control effectiveness evaluation model to monitor pest control indicators, non-target organism impact indicators, crop growth indicators, and environmental impact indicators; Based on farmland microecological simulation technology, predict the long-term ecological impact of prevention and control measures; Apply transfer learning mechanisms to optimize current strategies by leveraging historical prevention and control experience; Build adaptive learning and optimization models to achieve continuous evolution of the prevention and control system.

[0014] In a preferred embodiment, the adaptive learning and optimization model is based on a reinforcement learning framework, which treats the prevention and control process as a Markov decision process. The state space includes pest distribution, environmental parameters and crop growth status, the action space includes formula adjustment and application strategy modification, and the reward function is based on a comprehensive evaluation of prevention and control effect and resource consumption.

[0015] In a preferred embodiment, the 5G-based pest control system is used to perform a 5G-based pest control method, including: The data acquisition module includes a network of miniature bioactive sensors for collecting pest density, species, behavioral characteristics, resistance characteristics, and environmental parameters, and transmits the data in real time to edge computing nodes and a central data center via a 5G network. The intelligent decision-making module is used to implement a multi-objective evolutionary algorithm, optimize the formulation of biocontrol agents based on the collected data, and generate the best application strategy; The production execution module includes an intelligent synthesis unit for automatically synthesizing customized biocontrol agents based on optimized formulas; The precision delivery module is used to precisely deliver biocontrol agents at the optimal location, time, and dosage based on the calculation results of the environmentally adaptable microbial delivery algorithm. The effectiveness evaluation module is used to continuously collect data on prevention and control effects, analyze the effectiveness of prevention and control, and feed the data back to the intelligent decision-making module to optimize the formulation and application strategy, thereby achieving adaptive optimization of the system.

[0016] The beneficial effects of this invention are as follows: This invention enables customized synthesis and precise application of biocontrol agents. By combining a micro-bioactive sensor network and 5G network technology, it can collect pest information and environmental parameters in real time, optimize biocontrol agent formulations based on multi-objective evolutionary algorithms and preference-guided mechanisms, and produce customized biocontrol agents on-site through an intelligent synthesis unit. Compared with traditional general-purpose control agents, customized biocontrol agents are more targeted, can more effectively address specific pest species and environmental conditions, significantly improve control efficiency, and reduce the amount of control agents used.

[0017] This invention significantly improves the precision and efficiency of pest control. Its environmentally adaptive microbial delivery algorithm calculates the optimal delivery strategy based on pest distribution heatmaps and environmental parameters. By controlling precision delivery equipment via a 5G network, it achieves targeted and quantitative release of biological control agents. Compared to traditional large-scale uniform delivery methods, this invention concentrates control resources in high-risk areas while reducing unnecessary delivery to low-risk areas, greatly improving resource utilization efficiency and reducing waste.

[0018] By employing multi-objective optimization and preference-guided mechanisms, this invention reduces the negative impact on the ecological environment. In optimizing formulations, it fully considers the impact on non-target organisms and the ecological environment, especially in ecologically sensitive areas. By increasing the weight of environmental impact objectives, it can select more environmentally friendly formulations. Simultaneously, precision application technology reduces the overuse and drift of pesticides, lowering the risk of pollution to non-target areas.

[0019] This invention enhances the system's adaptability and self-improvement capabilities by continuously collecting data on prevention and control effectiveness. Utilizing multi-index evaluation models, microecological simulation technology, transfer learning mechanisms, and reinforcement learning frameworks, it constructs an adaptive learning and optimization model, enabling the entire system to continuously learn and improve from practice. With extended use, the system's prevention and control efficiency and resource utilization will continuously improve, and prevention and control strategies will be continuously optimized, forming a virtuous cycle of "becoming smarter with use."

[0020] This invention deeply integrates advanced technologies such as 5G, artificial intelligence, and edge computing with the field of biological control, constructing a complete intelligent pest control system. It not only solves problems in current biological control methods such as the inability to optimize formulations, lack of precision in application, and inability to adjust in real time, but also provides a technological innovation paradigm for agricultural pest control, leading the development of pest control technology towards intelligence and precision. The system's modular design gives it excellent scalability and adaptability, allowing for flexible adjustments based on different crops and pest types. Attached Figure Description

[0021] Figure 1 This is a flowchart of the 5G-based pest control method of the present invention. Detailed Implementation

[0022] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0023] At least one embodiment of the present invention discloses a 5G-based pest control method, such as... Figure 1 As shown, it includes the following steps: Step 1: Collect pest density, species, behavioral characteristics and environmental parameter data using a network of miniature bioactive sensors, and transmit the data to the processing system via a 5G network; Specifically, it includes: The network of miniature bioactive sensors includes various types of sensors deployed in different areas of farmland to comprehensively collect pest and environmental data.

[0024] The micro bioactive sensor network includes: Optical sensors are used to capture the species, quantity, and distribution of pests through image recognition technology; Acoustic sensors are used to monitor the sound wave signals emitted by pests. Chemical sensors are used to detect pheromones and volatile organic compounds released by pests; Environmental sensors are used to monitor environmental parameters such as temperature, humidity, light intensity, and rainfall. Soil sensors are used to collect data such as soil pH, nutrient content, and microbial activity.

[0025] These sensors employ a low-power design and are powered by solar energy or environmental energy harvesting technology to ensure long-term stable operation.

[0026] Step 2: Based on the pest and environmental data received by the processing system, a multi-objective evolutionary algorithm is used to optimize the formulation of biological control agents. The control problem is formalized into a multi-objective optimization problem that includes control efficiency, environmental impact, persistence and economic cost. A preference guidance mechanism is introduced to weigh the multiple objectives. This step details how the multi-objective evolutionary algorithm used in the system optimizes the formulation of biocontrol agents.

[0027] Problem modeling: The problem of optimizing the formulation of biocontrol agents is formalized into a multi-objective optimization problem, with decision variables including the concentration of each active ingredient, the proportion of auxiliary ingredients, and formulation parameters.

[0028] Optimization objectives include: The control efficiency target f1(θ) is calculated using the following formula: f1(θ)=∑(w i ×e i ×c i ) / ∑(w i ); Where θ is the formulation parameter vector, representing all parameters of the entire biocontrol agent formulation; e i c represents the unit efficacy index of the i-th component, indicating its ability to kill or inhibit the target pest; i Let w represent the concentration of the i-th component, indicating the proportion of that component in the entire formulation; i These are weighting coefficients used to adjust the importance of different components in the assessment of prevention and control efficiency; The environmental impact target f2(θ) is calculated using the following formula: f2(θ)=1-∑(p i ×ci ×t i ) / P max ; Where, p i t represents the environmental persistence index of the i-th component, indicating the duration of the component's presence in the environment; i The toxicity index indicates the degree of toxicity of the component to non-target organisms; c i P represents the concentration of the i-th component; max This is a standardization parameter used to standardize the environmental impact value to the range [0, 1]. The larger the value, the higher the environmental friendliness. The persistent objective f3(θ) is calculated using the following formula: f3(θ)=∑(d i ×c i ×s i ) / D max ; Where, d i c represents the degradation half-life of the i-th component, indicating the time required for the component to degrade by half under natural conditions; i s represents the concentration of the i-th component; i The stability coefficient represents the ability of a component to maintain its activity under different environmental conditions; D max This is a standardization parameter used to standardize persistence values ​​to the [0, 1] range; The economic cost objective f4(θ) is calculated using the following formula: f4(θ)=∑(c i ×p i )+P process ; Among them, c i p represents the concentration of the i-th component; i Let P be the unit cost of the i-th component, representing the market price or production cost of that component; process This refers to processing costs, including additional expenses incurred during the production process such as formula mixing, stabilization treatment, and quality control.

[0029] The constraints include: upper and lower limits of concentration for each component, compatibility constraints between components, stability constraints, and production feasibility constraints.

[0030] Algorithm initialization: The NSGA-III algorithm framework is adopted, and the initialization parameters include: population size N = 200, maximum number of iterations G = 500, and crossover probability p. c =0.9, initial mutation probability p m =0.1, number of reference points H=156 (uniform distribution in the four-dimensional target space). Real number encoding is used to represent the recipe parameters, and the initial population is generated by Latin hypercube sampling to ensure the diversity and uniform distribution of the initial solutions.

[0031] Population evolution: The evolutionary process proceeds in the following order: Parent selection: A binary tournament selection strategy is adopted; Genetic manipulation: Offspring are generated using simulated binary crossover (SBX) and polynomial mutation, with the mutation rate dynamically adjusted based on population diversity; Merge the parent and offspring populations; Non-dominated sorting: Dividing the merged population into different non-dominated layers; Reference point association: Calculate the perpendicular distance between each solution and the reference point; Environmental selection: The next generation of population is selected based on non-dominated ordination and reference point correlation.

[0032] Preference Guidance: Based on the needs of specific prevention and control scenarios, a preference guidance mechanism is introduced to dynamically adjust the weights of multiple objectives. The specific method involves constructing a comprehensive utility function. U(θ)=∑(w i ×f i (θ))+∑∑(w ij ×f i (θ)×f j (θ)); Where U(θ) represents the overall utility value of the formulation parameter vector θ; f i (θ) represents the value of the i-th optimization objective function, including prevention efficiency f1(θ), environmental impact f2(θ), persistence f3(θ), and economic cost f4(θ); w i w represents the weight coefficient of the i-th single objective, used to adjust the importance of that objective in the overall evaluation; ij The interaction weights between target i and target j are used to express the mutual influence relationships between different targets; ∑(w i ×f i (θ)) represents the weighted sum of each individual objective; ∑∑(w ij ×f i (θ)×f j (θ) represents a quadratic term that considers the interaction between objectives.

[0033] Different weighting configurations are used for different prevention and control scenarios: Ecologically sensitive areas: w2 = 0.4, w1 = 0.3, w3 = 0.2, w4 = 0.1; Pest outbreak periods: w1 = 0.5, w2 = 0.2, w3 = 0.2, w4 = 0.1; Long-term prevention and control area: w3=0.4, w1=0.3, w2=0.2, w4=0.1; Economic crops: w1 = 0.4, w4 = 0.3, w2 = 0.2, w3 = 0.1.

[0034] Solution set filtering: After running for G generations, the algorithm obtains a Pareto non-dominated solution set, typically containing 20 to 30 different recipient schemes. The system then further filters these schemes: Robustness analysis: Monte Carlo simulation was used to evaluate the performance stability of each formulation under small parameter fluctuations; Complexity assessment: Calculate the implementation difficulty index for each formula, including the difficulty of obtaining materials, the accuracy requirements of the formula, and the complexity of the production process; Overall score: Calculate the overall score for each recipe by combining utility function value, robustness, and implementation complexity.

[0035] Final decision: Based on the comprehensive score, the system selects the top three optimal formulas, conducts small-scale simulation tests, and finally determines the best formula scheme, generates detailed formula parameters, and transmits them to the intelligent synthesis unit for production.

[0036] In practical applications, the system, through the aforementioned multi-objective optimization method, can find the optimal balance point while considering multiple dimensions such as control efficiency, environmental impact, persistence, and economic cost, generating customized biocontrol agent formulations. Experiments have shown that, compared with traditional single-objective optimization methods, the formulations generated by this method have improved overall performance by more than 35%, especially achieving a significant breakthrough in balancing environmental friendliness and control efficiency.

[0037] Step 3: The optimized formula instructions are transmitted to the intelligent synthesis unit via the 5G network to produce customized biological control agents on-site. The central data center will digitally encode the formula parameters optimized by the multi-objective evolutionary algorithm to form a standardized production instruction data package. This data package contains comprehensive information such as the precise proportion of each bioactive component, the ratio of excipients, the selection of stabilizers, and the synthesis process parameters. Leveraging the high bandwidth and low latency characteristics of 5G networks, command data packets are transmitted in real time to intelligent synthesis units distributed in various prevention and control areas through encrypted channels. The transmission process employs end-to-end encryption technology and data integrity verification mechanisms to ensure the secure and reliable transmission of commands. After receiving the instruction, the intelligent synthesis unit immediately parses and verifies it to confirm its completeness and validity. At the same time, it checks the local resource status and assesses the execution capability. If any abnormality is found, it immediately reports to the central data center and requests supplementary instructions. Based on the precise parameters in the instruction, the intelligent synthesis unit controls key production parameters such as the amount of each component added, the order of addition, the reaction temperature, and the time. It uses microfluidic technology and a precision metering system to ensure nano-level addition accuracy and achieve precise proportioning of biocontrol agent components. During the production process, the intelligent synthesis unit has a variety of built-in sensors that monitor production parameters in real time, including key indicators such as pH value, temperature, humidity, redox potential, and microbial activity. These parameters are adjusted in real time through edge computing nodes to ensure that the production process meets the instructions and can make autonomous fine adjustments to maintain the best synthesis conditions when environmental conditions fluctuate. After synthesis, the intelligent synthesis unit performs rapid quality testing on the produced biocontrol agent, including analysis of active ingredient content, assessment of microbial activity, determination of physicochemical properties and prediction of stability, to ensure that the product meets the preset quality standards. The intelligent synthesis unit completes the on-site synthesis of customized biological control agents and feeds back the production completion status, product quality test data, and abnormal situations and resource consumption information during the production process to the central data center through the 5G network, providing data support for subsequent formula optimization and production efficiency improvement. Meanwhile, the intelligent synthesis unit will also perform necessary protective treatments on the synthesized biocontrol agents according to local environmental conditions, such as adding UV shielding agents, moisturizing agents, or microencapsulation, to enhance their stability and persistence in the field environment and further improve the control effect.

[0038] Step 4: Based on the pest and environmental data in the processing system, run the environmentally adaptive microbial delivery algorithm to generate a pest distribution heat map and calculate the optimal delivery strategy. Then, control the precision delivery device via the 5G network to release the produced biological control agent. Spatial data modeling: The system utilizes pest distribution data and environmental parameters collected by a micro-sensor network to construct a high-precision spatial data model. The specific steps are as follows: Data preprocessing: Outlier detection and correction are performed on the collected raw data, and noise is eliminated using methods such as moving average and Kalman filtering; Spatial interpolation: Using Kriging interpolation, discrete sampling point data is extended into a continuous spatial distribution function; Heat map generation: Based on the interpolation results, a pest density heat map H(x,y) is generated, where high-value areas represent areas with high pest density; Environmental parameter mapping: The same method is used to generate spatial distribution maps of environmental parameters such as temperature T(x,y), humidity M(x,y), and wind direction and speed W(x,y,t); Correlation analysis: Calculate the spatial correlation between pest distribution and environmental parameters to identify environmentally sensitive areas.

[0039] Deployment location optimization: Based on the constructed spatial data model, determine the optimal deployment location coordinate set P = {p1, p2, ..., p...} n}, where p1, p2, p nThese represent the coordinates of the 1st, 2nd, and nth optimal delivery locations, respectively, where n represents the number of optimal delivery location coordinates. The optimization process is as follows: Candidate point generation: The entire prevention and control area is divided into m×n grids, and the center of each grid is used as a candidate deployment point; Point importance calculation: For each candidate point p i Calculate its importance index I(p) i ): I(p i )=α×D(p i )+β×S(p i )+γ×E(p i ); Wherein, D(p) i ) represents the pest density factor, indicating the candidate point p. i The pest density level at the location, S(p i E(p) is the strategic location factor, which measures the importance of a point in the diffusion network and reflects its strategic value in the overall prevention and control network. i ) represents the environmental adaptability factor, indicating the degree of influence of environmental conditions at this point on the effectiveness of the control agent. α, β, and γ represent the weight coefficients of the pest density factor, strategy location factor, and environmental adaptability factor in the importance calculation, respectively.

[0040] Coverage effect modeling: Establish a coverage effect model for the delivery point p. i Its coverage effect on any point (x,y) in the region is as follows: C(p i (x,y)=e - ((xx i ) 2 +(yy i ) 2 ) / 2σ 2 ; Among them, C(p) i (x, y) represents the coverage effect model of the delivery point, (x i ,y i ) represents the coordinates of the application point, and σ is related to the diffusion characteristics of the pesticide; Greedy algorithm for point selection: The maximum marginal gain greedy algorithm is used to select k application points, and each time the point that maximizes the total coverage effect is selected; Refinement and optimization: Fine-tune the initially selected deployment points through a local search algorithm to further improve coverage efficiency.

[0041] Optimize delivery time: For each determined delivery location p i Determine the optimal delivery time t i The optimization process is as follows: Time window division: Divide 24 hours into several time windows, each with a length of 30 minutes; Time-based adaptability score: For each time window t, calculate its adaptability score A(p). i ,t): A(p i ,t)=w1×H(p i ,t)+w2×B(p i ,t)+w3×W(p i ,t); Among them, H(p) i ,t) represents the activity level of the pest, indicating the activity level at position p. i The activity level of pests at time t, B(p) i ,t) represents the activity of the biocontrol agent, indicating the activity at position p. i The effective activity level of the biocontrol agent at time t, W(p) i ,t) represents the suitability of environmental conditions, indicating the suitability at location p i The degree of influence of environmental conditions on the control effect at time t; w1, w2, and w3 represent the weight coefficients of pest activity, biocontrol agent activity, and environmental condition suitability in the adaptability score calculation, respectively. Optimal time selection: For each delivery point p i Select the time window with the highest adaptability score as the deployment time t. i ; Time conflict handling: Handling time scheduling conflicts between multiple delivery points to ensure the feasibility of delivery equipment scheduling.

[0042] Step 4, Dosage Calculation. For each dosing location p... i and time t i Calculate the optimal dosage d i The calculation process is as follows: Base dose calculation: Based on the pest density and distribution characteristics at this point, calculate the base dose d0(p i ); Environmental Adjustment: Adjusting the baseline dose based on environmental parameters. d1(p i )=d0(p i )×f T (T(p i ))×f M (M(p i )); Among them, f T and f M These are the correction functions for temperature and humidity, respectively, T(p) i ) indicates position p i The temperature value at that location, M(p) i ) indicates position p i The humidity value at that location; Wind direction and speed correction: Considering the impact of wind direction and speed on diffusion, the dosage is further corrected. d2(p i )=d1(p i )×f W (W(p i ,t i )) Among them, f W It is a correction function for wind force, W(p) i ,t i ) indicates at position p i and time t i Wind direction and wind speed values; Considerations for duration of effectiveness: Taking into account the degradation characteristics of the pesticide and the required control cycle, the final dosage was determined as follows: d i =d2(p i )×(1+λ×(T end -t i ) / T total ); Where λ is the persistence compensation coefficient, T end T represents the expected end time of the prevention and control cycle. total This refers to the total length of the prevention and control cycle.

[0043] Global optimization solution: Integrating delivery location, time, and dosage into a complete delivery strategy: S={(p1,t1,d1),(p2,t2,d2),...,(p n ,t n ,d n Global optimization is performed using simulated annealing algorithm: Objective function definition: F(S)=η1×E(S)-η2×C(S)-η3×R(S); Where E(S) represents the expected prevention and control effect, C(S) represents resource consumption, R(S) represents environmental risk, and η1, η2, and η3 are the weight coefficients representing the prevention and control effect, resource consumption, and environmental risk in the objective function, respectively. Initial solution generation: Generate an initial deployment strategy S0 based on the results of the preceding steps; Neighborhood operation definition: includes three types of operations: fine-tuning of delivery point, adjustment of time window, and modification of dosage; Annealing process: Initial temperature T0 = 100°C, cooling coefficient α = 0.95, final temperature T min =0.01, and perform L=100 neighborhood searches at each temperature; Optimal solution update: The global optimal solution S is retained during the iteration process. * The delivery strategy is used as the final output.

[0044] After the optimization solution is completed, the system will assign the optimal deployment strategy S. * The data is transmitted to the precision delivery device module for actual delivery. During delivery, the system monitors environmental changes in real time via the 5G network and makes fine-tuning adjustments to the strategy as needed to ensure the accuracy and adaptability of the delivery process.

[0045] Experimental results show that, compared with the traditional uniform application method, the use of this environmentally adaptive microbial application algorithm can reduce the amount of control agent by more than 40% while achieving the same control effect, and increase the effective coverage rate of control to more than 95%, significantly improving resource utilization efficiency and control precision.

[0046] Step 5: Continuously collect data on prevention and control effects through a network of micro-bioactive sensors, and feed it back to the central data center via a 5G network. Optimize the formulation and application strategy based on the feedback data.

[0047] This step describes in detail the reinforcement learning-based adaptive optimization model used in the system to continuously improve the prevention and control strategy.

[0048] Markov decision process modeling: formalizing the prevention and control process as a quintuple of MDP (S, A, P, R, γ): State space S: includes the distribution state of pests s pest Environmental parameter status s env and crop growth status crop The combined representation is s = [s pest 's env 's crop ]; Action Space A: Includes recipe adjustment action a formula and delivery strategy action a deploy The combined expression is a = [a formula 'a deploy ]; State transition probability P: describes the probability that the system will transition from state s to state s′ after performing action a, P(s′|s,a), which is estimated through historical data and farmland micro-ecological model; Reward function R: R(s,a,s′) is defined as w1×E(s′)-w2×C(a)-w3×I(s′), where E(s′) is the prevention and control effect score, C(a) is the resource consumption, I(s′) is the environmental impact, and w1, w2, and w3 represent the weight coefficients of prevention and control effect, resource consumption, and environmental impact in the reward calculation, respectively. Discount factor γ: set to 0.9 to balance short-term and long-term returns.

[0049] Deep Q-Network Architecture Design: A Double Deep Q-Network (DQN) architecture is adopted to reduce the overestimation problem of Q-value estimation. Network structure: The input layer receives the state vector s, and the output layer generates the Q-value estimate for each action a; Hidden layer design: It contains 3 fully connected layers, each with 256, 128 and 64 neurons respectively, using the ReLU activation function; Experience replay buffer: capacity is 10000, storing transferred samples (s,a,r,s′); Target network update: Update once every 100 steps, using a soft update method: θ′←τθ+(1-τ)θ′, where τ=0.01.

[0050] Priority experience replay mechanism: To improve learning efficiency, priority experience replay is implemented. Priority calculation: p i =|δ i |+ε, where δ i For TD error, ε = 0.01 is a small constant; Sampling probability: Where α = 0.6 controls the degree of priority; Importance sampling weight: w i =(N·P(i)) -β , where β increases linearly from 0.4 to 1.0.

[0051] Multi-objective reinforcement learning extension: Extending the single reward function to a multi-objective form allows for a more comprehensive evaluation of prevention strategies. Vectorized reward: R(s,a,s′)=[r 效果 'r 成本 'r 环境 ]; Linear preference setting: w = [w 效果 'w 成本 'w 环境 Adjustments will be made dynamically based on prevention and control needs. Scalar reward calculation: r = w·R(s,a,s′).

[0052] Transfer learning mechanism: Utilizing historical prevention and control experience and knowledge of similar scenarios to accelerate the learning process in the current scenario. Source task identification: Identify source tasks similar to the current prevention and control scenario from historical databases; Knowledge extraction: Extracting relevant features, network parameters, and value functions from the source task; Parameter transfer: The extracted knowledge is used to initialize the network parameters of the current task; Fine-tuning and optimization: Further fine-tune the network parameters for the current task.

[0053] Farmland micro-ecology simulation integration: Integrating farmland micro-ecology simulation technology with reinforcement learning frameworks to improve model prediction accuracy. Simulated environment construction: Based on ecological principles, a simulated micro-ecological environment for farmland is constructed, including pest population dynamics, natural enemy interactions, and environmental impacts; Model pre-training: Pre-training the reinforcement learning model in a simulated environment to accumulate initial experience; Environmental calibration: Continuously calibrate simulated environmental parameters based on actual monitoring data; Parallel policy evaluation: Evaluate multiple candidate policies in parallel in a simulated environment to reduce the actual trial and error cost.

[0054] Online learning and continuous optimization: The system continuously learns from actual prevention and control processes, constantly improving prevention and control strategies. Data collection: Real-time collection of status transition and reward information during the prevention and control process; Incremental learning: updating model parameters using newly collected data; Exploration mechanism: An ε-greedy strategy is adopted, with the ε value gradually decreasing from 0.3 to 0.05; Strategy evaluation: Regularly evaluate the performance of the current strategy and compare it with the best historical strategy; Model update: When the cumulative performance improvement exceeds the threshold, update the strategy model in the production system.

[0055] Through the aforementioned reinforcement learning-based adaptive pest control strategy optimization system, this invention achieves a self-improving capability that becomes "smarter with use." In practical applications, after six months of continuous learning, the system's pest control decision accuracy increased from an initial 75% to over 95%, resource utilization efficiency improved by 40%, and the adaptability of the pest control strategy to environmental changes was significantly enhanced. This system can continuously optimize formulations and application strategies based on feedback from pest control effectiveness and changes in environmental conditions, providing strong support for long-term sustainable pest control.

[0056] System application experiments and effect verification: This embodiment details the application experiments and effect verification of the system of the present invention in a real farmland environment. Three different types of experimental fields were selected, representing different crop types, pest types, and environmental conditions, and the experimental period was one complete growing season (150 days).

[0057] Experimental field A: Wheat field, covering an area of ​​50 hectares, with the main pest being wheat aphids, located in the North China Plain, where the climate is relatively dry.

[0058] Experimental field B: a paddy field with an area of ​​30 hectares, where the main pest is the rice planthopper, located in the Jiangnan region with a humid climate.

[0059] Experimental field C: Vegetable greenhouse, covering an area of ​​10 hectares, where multiple pests occur together, belonging to the facility agriculture environment.

[0060] Each experimental field was divided into two parts: one part used the 5G-based pest control system of this invention (experimental group), and the other part used traditional biological control methods (control group). The system deployment and verification process is as follows: System Deployment: Deploy a complete 5G-based pest control system in the experimental area: Sensor network deployment: Depending on the size and shape of the field, install 6 to 10 miniature bioactivity sensor nodes per hectare to cover key monitoring points; Edge computing node deployment: One edge computing unit is set up every 10-15 hectares to be responsible for data preprocessing; 5G network coverage: Cooperate with local telecom operators to ensure full 5G signal coverage in the trial area; Deployment of intelligent synthesis stations: One intelligent synthesis station is set up every 15-20 hectares, depending on the distribution of the fields; Precision delivery equipment preparation: Configure ground-based automatic navigation delivery vehicles and drone delivery systems, and allocate them reasonably according to the size of the fields.

[0061] System debugging: After deployment, a two-week system debugging period will be conducted. Sensor calibration: Various sensors are calibrated using standard samples to ensure data accuracy; Network testing: Test the coverage quality, latency, and bandwidth of the 5G network to ensure that system requirements are met; Algorithm initialization: Input historical pest occurrence patterns and environmental data to initialize the system algorithm model; Small-scale testing: Select a small area to conduct full-process system testing to verify the collaborative working ability of each module; Parameter optimization: Adjust system parameters based on test results to optimize performance.

[0062] Evaluation of control effectiveness: The effectiveness of control measures should be evaluated regularly throughout the growing season. Pest density monitoring: Weekly statistics on pest population density and distribution; Natural enemy survival rate: assesses the survival of beneficial organisms during biological control; Crop growth indicators: monitoring crop growth, yield, and quality indicators; Resource consumption: Record the amount of pesticides used, equipment operating time, and energy consumption; Environmental impact: Monitoring soil microbial diversity, water quality, and non-target biological impacts.

[0063] The experimental results are analyzed as follows: Pest control effect: Experimental field A (wheat field): The average aphid density in the experimental group was 65% lower than that in the control group, and the density rebound period was extended by 2.5 times; Experimental field B (rice field): The average density of rice planthoppers in the experimental group was 58% lower than that in the control group, and the population growth rate was reduced by 70%. Experimental field C (vegetable greenhouse): The integrated pest management index of the experimental group was 75% higher than that of the control group.

[0064] Resource consumption comparison: Dosage of control agent: The experimental group was reduced by an average of 48% compared with the control group; Labor input: The experimental group had 85% less labor input than the control group; Energy consumption: Although the system itself requires energy, the energy consumption per unit of prevention and control effect was 35% lower in the experimental group than in the control group.

[0065] Environmental friendliness assessment: Non-target organism impacts: The negative impacts on pollinating insects, natural enemies, and soil microorganisms in the experimental group were reduced by 83% compared to the control group; Biodiversity index: The Shannon diversity index in the experimental area was 0.8-1.2 higher than that in the control area; Residue assessment: No significant residues were detected in the soil and crops of the experimental group fields.

[0066] Crop yield and quality: Increased yield: The experimental group yielded an average of 15-22% more than the control group; Quality improvement: The pesticide residue detection rate of crops in the experimental group was 95% lower than that in the control group, and the quality score was improved by 18%.

[0067] Systematic learning and evolutionary capabilities: Decision accuracy: The system gradually improved from an initial 72% to 94%; Resource utilization efficiency: As the system usage time increases, the resource input required per unit of control effect continues to decrease, and by the end of the 150-day trial period, it is 45% lower than the initial level. Adaptability: When faced with sudden weather changes or pest outbreaks, the system can automatically adjust its strategy within 24 hours to maintain the control effect.

[0068] Economic benefit analysis: Initial investment: The initial cost of system deployment is 280% higher than that of traditional methods; Operating costs: The system's daily operating costs are 60% lower than traditional methods; Payback period: Through increased production and cost savings, the system investment payback period is 2 to 3 growing seasons; Long-term benefits: Over a 5-year period, the overall economic benefits of the system are 135% higher than those of traditional methods.

[0069] Experimental Conclusions: In three different types of experimental fields, the 5G-based pest control system of this invention demonstrated significant technical advantages, comprehensively surpassing traditional methods in terms of pest control effectiveness, resource utilization efficiency, environmental friendliness, and adaptability. In particular, the system's self-learning and continuous optimization capabilities enabled its effectiveness to continuously improve over long-term use, providing a new technological path for modern agricultural pest control. Even considering the relatively high initial investment, from a medium- to long-term economic perspective, this system still possesses significant economic feasibility and promotional value.

[0070] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A 5G-based pest control method, characterized in that, Includes the following steps: A network of miniature bioactive sensors is used to collect data on pest density, species, behavioral characteristics, and environmental parameters, and the data is transmitted to the processing system via a 5G network. Based on pest and environmental data received by the processing system, a multi-objective evolutionary algorithm is used to optimize the formulation of biological control agents. The control problem is formalized into a multi-objective optimization problem that includes control efficiency, environmental impact, persistence and economic cost. A preference guidance mechanism is introduced to weigh the multiple objectives. The optimized formula instructions are transmitted to the intelligent synthesis unit via 5G network, enabling the on-site production of customized biological control agents; Based on pest and environmental data in the processing system, an environmentally adaptive microbial delivery algorithm is run to generate a pest distribution heat map and calculate the optimal delivery strategy. The produced biological control agent is then released by a precision delivery device controlled by a 5G network. The system continuously collects data on prevention and control effects through a network of miniature bioactive sensors, and feeds this data back to the central data center via a 5G network. Based on the feedback data, the formulation and application strategy are optimized.

2. The 5G-based pest control method according to claim 1, characterized in that, The preference guidance mechanism dynamically adjusts the weight of a single objective and the interaction weight between objectives according to the application scenario. The environmentally adaptive microbial delivery algorithm considers three key dimensions: delivery location, time, and dosage, to calculate the globally optimal delivery strategy.

3. The 5G-based pest control method according to claim 1, characterized in that, The micro bioactive sensor network includes optical sensors, acoustic sensors, chemical sensors, environmental sensors, and soil sensors, used to collect basic information about pests, behavioral characteristic data, resistance characteristics, and environmental correlation data; the collected data is preprocessed through edge computing nodes and then transmitted to the central data center for in-depth analysis.

4. The 5G-based pest control method according to claim 1, characterized in that, The multi-objective evolutionary algorithm is a non-dominated sorting genetic algorithm, including: Formula parameters are represented using real number encoding; Individual selection is performed by combining non-dominated ranking and reference point methods; An adaptive mutation operation is introduced, and the mutation rate is dynamically adjusted according to the population diversity. Introducing a local search mechanism in the later stages of evolution improves the accuracy of the solution; Output a set of Pareto undominated solutions, each solution representing an optimal recipe under multiple objective trade-offs.

5. The 5G-based pest control method according to claim 1, characterized in that, The preference guidance mechanism comprehensively evaluates multiple objectives through a comprehensive utility function and dynamically adjusts the weights in the following manner: For ecologically sensitive areas, increase the weight of environmental impact targets; The weighting of the goal of improving control efficiency during pest outbreak periods; For areas requiring long-term prevention and control, increase the weight of persistent objectives; For cash crops, increase the weighting of cost and efficiency objectives.

6. The 5G-based pest control method according to claim 1, characterized in that, The algorithm for deploying environmentally adaptive microorganisms includes: Construct a spatial data model of pest distribution and generate a heat map of pest distribution; Determine the coordinate set of release locations, prioritizing areas with high pest density and key diffusion nodes; Determine the set of release time points and select the best release time based on pest activity patterns and environmental factors; Calculate the optimal dosage for each location, and dynamically adjust the dosage based on pest density and environmental conditions; The global optimal deployment strategy is solved using the simulated annealing optimization method.

7. The 5G-based pest control method according to claim 1, characterized in that, The intelligent synthesis unit produces customized biological control agents according to the following steps: The optimized formula parameters are converted into specific executable instructions, including the precise amount of each component to be added, the order of addition, reaction time and temperature control parameters; Accurately measure each component according to instructions, and mix the components in the specified order and under the specified conditions; Add stabilizers and perform necessary physical treatments to ensure the stability of the control agent; By using built-in sensors to monitor key parameters of the synthesis process in real time, we ensure that product quality meets expectations. The synthesized control agents are evaluated from multiple dimensions, including physical properties, chemical properties, and biological activity, and the test results are fed back to the system center.

8. The 5G-based pest control method according to claim 1, characterized in that, The continuously collected data on prevention and control effectiveness includes: Construct a multi-indicator prevention and control effectiveness evaluation model to monitor pest control indicators, non-target organism impact indicators, crop growth indicators, and environmental impact indicators; Based on farmland microecological simulation technology, predict the long-term ecological impact of prevention and control measures; Apply transfer learning mechanisms to optimize current strategies by leveraging historical prevention and control experience; Build adaptive learning and optimization models to achieve continuous evolution of the prevention and control system.

9. The 5G-based pest control method according to claim 7, characterized in that, The adaptive learning and optimization model is based on a reinforcement learning framework, which treats the prevention and control process as a Markov decision process. The state space includes pest distribution, environmental parameters and crop growth status, the action space includes formula adjustment and application strategy modification, and the reward function is based on a comprehensive evaluation of prevention and control effect and resource consumption.

10. A 5G-based pest control system, used to execute the 5G-based pest control method according to any one of claims 1-9, characterized in that, include: The data acquisition module includes a network of miniature bioactive sensors for collecting pest density, species, behavioral characteristics, resistance characteristics, and environmental parameters, and transmits the data in real time to edge computing nodes and a central data center via a 5G network. The intelligent decision-making module is used to implement a multi-objective evolutionary algorithm, optimize the formulation of biocontrol agents based on the collected data, and generate the best application strategy; The production execution module includes an intelligent synthesis unit for automatically synthesizing customized biocontrol agents based on optimized formulas; The precision delivery module is used to precisely deliver biocontrol agents at the optimal location, time, and dosage based on the calculation results of the environmentally adaptable microbial delivery algorithm. The effectiveness evaluation module is used to continuously collect data on prevention and control effects, analyze the effectiveness of prevention and control, and feed the data back to the intelligent decision-making module to optimize the formulation and application strategy, thereby achieving adaptive optimization of the system.

Images