Method and system for intelligently monitoring and accurately preventing and treating plant diseases and insect pests in traditional Chinese medicinal material standard base

Abstract

The invention discloses a traditional Chinese medicinal material standard base pest and disease damage intelligent monitoring and precise control method and control system, and relates to the technical field of agricultural intelligence. The Internet of Things technology, the block chain and the artificial intelligence algorithm are integrated to improve monitoring accuracy and prevention timeliness, environment and plant states are acquired through the sensor and image data, safe storage of data is ensured by using the block chain, early recognition of diseases and pests is performed by using the convolutional neural network, and the accuracy of disease and pest control is improved. A disease risk index is calculated in combination with environmental parameters, a reinforcement learning algorithm dynamically optimizes a control strategy, control measures are automatically executed through Internet of Things equipment, an execution result is recorded on a block chain, in addition, the strategy is continuously adjusted according to a control effect evaluation result, intelligent and precise disease and pest management is realized, resource consumption is effectively reduced, and the economic benefit is improved. The traditional Chinese medicinal material production safety and efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural intelligence, in particular to a method and system for intelligent monitoring and precise prevention and treatment of diseases and pests in a standard base of traditional Chinese medicinal materials. BACKGROUND

[0002] Traditional Chinese medicinal material planting is an important foundation for ensuring the sustainable development of traditional Chinese medicine industry, and the prevention and treatment of diseases and pests thereof is directly related to the quality and yield of medicinal materials, and is of key significance to promoting agricultural modernization and the Healthy China strategy. However, the existing methods for preventing and treating diseases and pests rely on manual experience and traditional monitoring means, and generally have problems such as insufficient data sharing, low monitoring accuracy, and delayed prevention and treatment decision-making, which leads to difficulty in optimizing the prevention and treatment effect and serious waste of resources, especially in terms of data credibility, early disease identification, and dynamic prevention and treatment strategy formulation, the traditional methods are difficult to meet the high standard requirements of precision agriculture.

[0003] The core challenges are concentrated in the following technical factors: first, there is a lack of efficient mechanism for the credible transmission and storage of environmental data and pest information, which affects data sharing and collaborative analysis; second, the existing monitoring technology cannot capture the early disease characteristics of traditional Chinese medicinal materials, which limits the precision identification capability; and finally, the prevention and treatment decision lacks a self-adaptive optimization mechanism, and cannot dynamically adjust the strategy according to real-time data, which leads to low efficiency of disease and pest prevention and treatment, and makes it difficult to realize the intelligentization of the whole process. SUMMARY

[0004] The present application relates to the technical field of agricultural intelligence, in particular to a method and system for intelligent monitoring and precise prevention and treatment of diseases and pests in a standard base of traditional Chinese medicinal materials.

[0005] The object of the present application can be achieved by the following technical solutions:

[0006] The present application provides a method for intelligent monitoring and precise prevention and treatment of diseases and pests in a standard base of traditional Chinese medicinal materials, comprising the following steps:

[0007] Plant image data is extracted from the credible data block, a convolutional neural network model is used to extract features from the image, and a pre-established early disease and pest feature library is combined. When the similarity of the extracted features and the features in the library exceeds a preset threshold T1, early disease is determined, and an early disease and pest identification result is obtained.

[0008] According to the early identification result, the temperature, humidity and other parameters in the environmental data are fused, a multi-modal data fusion algorithm is used, and a comprehensive disease risk index is calculated by a weighted average method to obtain a disease risk assessment result.

[0009] According to the disease risk assessment result, a reinforcement learning algorithm is used to construct a state space, an action space and a reward function, and a dynamically optimized prevention and treatment strategy is obtained by Q-learning iterative updating strategy.

[0010] extracting specific action instructions from the dynamically optimized prevention and control strategy, controlling agricultural machinery to perform spraying pesticides or adjusting irrigation through Internet of Things devices, generating an execution log, storing it in a blockchain ledger, and obtaining an implementation record of the prevention and control measures;

[0011] obtaining the implementation record and real-time environmental data, using a time series analysis algorithm to calculate the change trend of the disease risk index, and if the change trend shows that the risk index decreases and is lower than a preset threshold T2, determining that the prevention and control strategy is effective, and obtaining a prevention and control effect evaluation result;

[0012] According to the prevention and control effect evaluation result, updating the reward function of the reinforcement learning algorithm, adjusting the reward weight to improve the preference for low resource consumption strategies, generating an updated dynamic prevention and control strategy, storing it in a distributed database, and obtaining an adaptive prevention and control strategy optimization result;

[0013] extracting the latest strategy from the adaptive prevention and control strategy optimization result, combining the real-time data processing module, and through the closed-loop process of data collection, risk assessment and strategy optimization, generating continuously updated disease and pest control instructions, storing them in a blockchain ledger, and obtaining the running result of the whole-process intelligent prevention and control system.

[0014] Further, the early disease and pest identification result is obtained, specifically including:

[0015] obtaining plant image data from the trusted data block, using preprocessing techniques to denoise and standardize the image to obtain first image data, and then using a convolutional neural network to extract features from the first image data to generate an image feature set;

[0016] obtaining a reference feature set from a pre-established early disease and pest feature library, calculating the similarity between the image feature set and the reference feature set to obtain a feature similarity value, and when the feature similarity value exceeds a preset threshold T1, determining that it is an early disease, and generating an early disease and pest identification result;

[0017] According to the early disease and pest identification result, using a classification algorithm to distinguish the disease type, obtaining a disease type label, and through the disease type label, obtaining the corresponding disease distribution characteristics from a pre-established disease knowledge base to generate a disease distribution characteristic set;

[0018] According to the disease distribution characteristic set, using a clustering algorithm to perform zonal analysis on the disease distribution to obtain a disease distribution area result.

[0019] Further, the disease risk assessment result is obtained, specifically including:

[0020] obtaining temperature and humidity parameters in the environmental data, and then fusing the standardized environmental data set and the image feature data set to generate a fused feature data set using a multi-modal data fusion algorithm;

[0021] The comprehensive disease risk index is calculated by a weighted average method to obtain a risk index value. When the risk index value exceeds a preset threshold, it is determined that there is a high disease risk, and a high-risk assessment result is generated. When the risk index value is lower than the preset threshold, it is determined that there is a low disease risk, and a low-risk assessment result is generated.

[0022] According to the high-risk assessment result or the low-risk assessment result, corresponding disease risk assessment report data is generated.

[0023] Further, a dynamically optimized prevention and control strategy is obtained, which specifically includes:

[0024] The environmental data and the disease risk assessment result are obtained to construct a state space, which includes temperature, humidity, soil parameters and risk index, and the dimension of the state space is determined. Then, the action space is defined through the state space, which includes the types and doses of the sprayed pesticides, the adjustment of irrigation frequency and water quantity, and the action space set is obtained.

[0025] According to the prevention and control effect and the resource consumption, a reward function is constructed. When the prevention and control effect is higher than a preset threshold and the resource consumption is lower than a preset threshold, the reward value is updated positively to obtain a reward function model.

[0026] The Q-learning algorithm is used to initialize the Q value table. The Q value is iteratively updated for the state space and the action space to obtain a dynamic strategy. Then, the optimal action combination of the current state is extracted from the dynamic strategy to generate a prevention and control scheme including spraying pesticides and adjusting irrigation.

[0027] After the execution of the prevention and control scheme, the environmental data and the disease risk index are obtained. If the risk index is lower than a preset threshold, the current scheme is maintained. Otherwise, the state space is re-entered for updating to obtain a new prevention and control scheme.

[0028] Through multiple iterations, the prevention and control effect and the resource consumption of each prevention and control scheme are recorded, and the reward function is updated to obtain an optimized dynamic prevention and control scheme.

[0029] Further, the implementation record of the prevention and control measures is obtained, which specifically includes:

[0030] The action instructions are extracted from the dynamically optimized prevention and control strategy, and the natural language processing technology is used to analyze the strategy text to identify the instruction content of the sprayed pesticides or the adjusted irrigation to obtain an action instruction set.

[0031] When the action instruction set includes the sprayed pesticide instruction, the control signal is sent to the agricultural machinery through the Internet of Things device to obtain the mechanical operation state and determine the execution of the sprayed pesticide task. When the action instruction set includes the adjusted irrigation instruction, the control signal is sent to the agricultural machinery through the Internet of Things device to obtain the irrigation parameter adjustment data and determine the execution of the irrigation task.

[0032] According to the spraying pesticide task performed by the agricultural machine, a log generation algorithm is adopted to record pesticide types, spraying time and area information to obtain a spraying execution log, and according to the adjustment irrigation task performed by the agricultural machine, a log generation algorithm is adopted to record irrigation time, water quantity and area information to obtain an irrigation execution log;

[0033] Through a blockchain ledger interface, the spraying execution log and the irrigation execution log are uploaded to the blockchain ledger to obtain a storage transaction hash value, and the log storage is determined to be completed, and the stored execution log is extracted from the blockchain ledger, a data integration technology is adopted to generate a prevention and control measure implementation record containing a timestamp, a task type and an area to obtain a final implementation record.

[0034] Further, a prevention and control effect evaluation result is obtained, specifically including:

[0035] The implementation record data and real-time environment data are obtained, and through data cleaning and formatting processing, a standardized time series data set is generated, and a time series analysis algorithm is adopted to process the time series data set to calculate a disease risk index to obtain a risk index sequence;

[0036] The risk index sequence is analyzed by a sliding window method to calculate the change trend of the risk index to obtain a trend feature sequence, and when the trend feature sequence shows that the risk index continuously decreases, a comparison method is used to compare with a preset threshold T2 to determine whether it is lower than the threshold to obtain a threshold comparison result;

[0037] According to the threshold comparison result, a logical judgment rule is adopted, and when the risk index is lower than the preset threshold T2, it is determined that the prevention and control strategy is effective to obtain a strategy effectiveness identifier;

[0038] The strategy effectiveness identifier and the risk index sequence are summarized to generate prevention and control effect evaluation data to obtain an evaluation result data set, and a visualization technology is adopted to process the evaluation result data set to generate a change trend chart to obtain dynamic display data of the prevention and control effect.

[0039] Further, an adaptive prevention and control strategy optimization result is obtained, specifically including:

[0040] Key indicator data is obtained from the prevention and control effect evaluation result, a statistical analysis method is adopted to calculate an effect score and a resource consumption score to obtain an evaluation index set, and when the effect score is lower than a preset threshold, a gradient descent algorithm is used to optimize a reward function of a reinforcement learning model to adjust a reward weight to obtain an updated reward function;

[0041] According to the updated reward function, a reinforcement learning algorithm is used to generate a dynamic prevention and control strategy, determine the adaptive parameter set of the strategy, and then extract the resource consumption features from the dynamic prevention and control strategy to determine whether the low resource consumption condition is met. When the condition is met, the candidate strategy set is stored in the distributed database to obtain the candidate strategy set.

[0042] The candidate strategy set is verified in parallel using a distributed computing framework to obtain the performance indicators of the verified strategies, determine the preferred strategy set, and update the initial parameters of the reinforcement learning model using the preferred strategy set to generate the final adaptive prevention and control strategy, which is stored in the distributed database to obtain the optimized strategy result.

[0043] The optimized strategy result is extracted from the distributed database, and a consistency checking algorithm is used to verify the data integrity to obtain the final prevention and control strategy output.

[0044] Further, before extracting the plant image data from the trusted data block, it further includes:

[0045] Through the sensor network and the Internet of Things device, environmental data such as temperature, humidity, soil nutrients, and light intensity are obtained from the traditional Chinese medicinal material planting environment, combined with the plant image data captured by the camera, a multi-modal data set containing a timestamp is generated, and stored in the distributed database to obtain the data set of the environment and plant state. Through blockchain technology, the data set is encrypted, and the data source and integrity are verified through a smart contract to generate a data block containing a hash value, which is stored in a distributed ledger to obtain a trusted data transmission and storage result.

[0046] Further, the trusted data transmission and storage result specifically includes:

[0047] The initial data set is encrypted through blockchain technology to generate an encrypted data set. If the encrypted data set passes the preset encryption strength verification, the data source and integrity are verified through a smart contract to obtain a verification result. According to the verification result, a data block containing a hash value is generated.

[0048] The data block is stored in the distributed ledger to obtain a storage address, and the data block in the storage address is verified using a hash value to determine the data integrity. If the data integrity passes the verification, a trusted data transmission record is obtained, and a trusted data storage result is generated according to the transmission record.

[0049] The present application provides a traditional Chinese medicinal material standard base pest intelligent monitoring and precision prevention and control system for realizing a traditional Chinese medicinal material standard base pest intelligent monitoring and precision prevention and control method, comprising:

[0050] The data collection and verification module uses sensor networks and IoT devices to collect key environmental data and plant images from the TCM planting environment, creates a multimodal dataset with timestamps, and stores it in a distributed database. Blockchain technology is used for encryption and integrity verification, generating data blocks containing hash values.

[0051] The early identification module for pests and diseases uses a convolutional neural network model to extract features from plant images and compares them with a database of early pest and disease features to identify early lesions. It then uses a classification algorithm to distinguish lesion types and obtains corresponding disease distribution features from a disease knowledge base to analyze disease distribution areas.

[0052] The disease risk assessment module integrates environmental data and image feature data, uses multimodal data fusion algorithm to calculate the comprehensive disease risk index, and generates a disease risk assessment report based on the risk assessment results.

[0053] The prevention and control strategy generation module uses reinforcement learning algorithms to construct state space, action space, and reward function, and iteratively updates the strategy through Q-learning to generate dynamically optimized prevention and control strategies;

[0054] The prevention and control measures implementation and recording module extracts specific action instructions from the dynamically optimized prevention and control strategies, controls agricultural machinery through IoT devices to perform operations such as spraying pesticides or adjusting irrigation, generates execution logs, stores them in the blockchain ledger, and records the implementation of prevention and control measures;

[0055] The effect evaluation and strategy optimization module obtains implementation records and real-time environmental data, uses a time series analysis algorithm to calculate the changing trend of the disease risk index, updates the reward function of the reinforcement learning algorithm based on the evaluation results, and generates adaptive control strategy optimization results;

[0056] The intelligent pest control system operation module extracts the latest strategy from the adaptive prevention and control strategy optimization results, combines it with the real-time data processing module, and generates continuously updated pest control instructions through a closed-loop process of cyclically executing data collection, risk assessment and strategy optimization, which are stored in the blockchain ledger.

[0057] The beneficial effects of the present invention are:

[0058] This invention integrates sensor networks and IoT devices to enable real-time collection of multimodal data on the TCM planting environment, including environmental parameters and plant images. This data is securely stored in a distributed database and encrypted and integrity verified using blockchain technology, ensuring data security and reliability. This process addresses the challenges of insufficient data sharing and low monitoring accuracy in traditional monitoring methods, laying a solid data foundation for the early identification and precise prevention of pests and diseases.

[0059] The collected plant images are subjected to deep feature extraction by using a convolutional neural network, and are compared and analyzed with an early disease and pest feature library, when the feature similarity exceeds a preset threshold, early diseases can be accurately identified, the disease types are distinguished through a classification algorithm, and the disease distribution area is analyzed combined with a disease knowledge base, the early identification ability of diseases and pests is significantly improved, the prevention and control measures can be intervened earlier, and the negative influence of diseases and pests on the quality and yield of traditional Chinese medicinal materials is effectively reduced.

[0060] The application adopts a reinforcement learning algorithm, dynamically optimizes the prevention and control strategy according to real-time monitoring data and disease and pest risk assessment results, automatically executes the optimal prevention and control measures such as spraying pesticides or adjusting irrigation through Internet of Things equipment, and records the execution results in a blockchain account book, and can continuously adjust and optimize the prevention and control strategy according to the prevention and control effect evaluation results, realizes the intelligentization and precision of the prevention and control decision, not only improves the prevention and control efficiency, but also reduces the resource waste, and provides strong technical support for the intelligent management of diseases and pests of traditional Chinese medicinal materials. BRIEF DESCRIPTION OF DRAWINGS

[0061] In order to better understand and implement, the technical scheme of the present application is described in detail below in combination with the drawings.

[0062] Figure 1 The process schematic diagram of the intelligent monitoring and precision prevention and control method of diseases and pests of traditional Chinese medicinal materials provided for the embodiment 1 of the present application is shown in the figure.

[0063] Figure 2 The process schematic diagram of the intelligent monitoring and precision prevention and control method of diseases and pests of traditional Chinese medicinal materials provided for the embodiment 1 of the present application is shown in the figure.

[0064] Figure 3 The process schematic diagram of the intelligent monitoring and precision prevention and control method of diseases and pests of traditional Chinese medicinal materials provided for the embodiment 1 of the present application is shown in the figure.

[0065] Figure 4 The structure schematic diagram of the intelligent monitoring and precision prevention and control system of diseases and pests of traditional Chinese medicinal materials provided for the embodiment 2 of the present application is shown in the figure. DETAILED DESCRIPTION

[0066] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purposes, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application. On the contrary, they are only examples of methods and systems consistent with some aspects of the present application as detailed in the appended claims.

[0067] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in this application and the appended claims, the singular forms "a," "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that the term "and / or," as used herein, refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0068] The specific embodiments according to the present application, features and effects thereof will be described in detail below with reference to the accompanying drawings and preferred embodiments.

[0069] Embodiment 1

[0070] Please refer to Figures 1-3 The embodiment provides a method for intelligent monitoring and precise prevention of plant diseases and insect pests in a traditional Chinese medicine material standard base, comprising the following steps:

[0071] S1, extracting plant image data from a trusted data block, using a convolutional neural network model to extract features from the image, and combining a pre-established early disease and insect pest feature library, when the similarity of the extracted features and the features in the library exceeds a preset threshold T1 (T1 = 0.85), it is determined that there is early disease, and an early disease and insect pest identification result is obtained;

[0072] Further, in step S1, the early disease and insect pest identification result is obtained, specifically comprising:

[0073] Obtaining plant image data from a trusted data block, using preprocessing techniques to denoise and standardize the image to obtain first image data, and then using a convolutional neural network to extract features from the first image data to generate an image feature set;

[0074] Obtaining a reference feature set from a pre-established early disease and insect pest feature library, calculating the similarity of the image feature set and the reference feature set to obtain a feature similarity value, and when the feature similarity value exceeds a preset threshold T1, determining that there is early disease, and generating an early disease and insect pest identification result;

[0075] According to the early disease and insect pest identification result, using a classification algorithm to distinguish the disease type, obtaining a disease type label, and through the disease type label, obtaining corresponding disease distribution characteristics from a pre-established disease knowledge base to generate a disease distribution characteristic set;

[0076] According to the disease distribution characteristic set, using a clustering algorithm to analyze the disease distribution, and obtaining a disease distribution area result.

[0077] For example, in the three-leaf planting area of a traditional Chinese medicine standard base, multispectral imaging equipment and high-definition cameras are deployed in the base to collect three-leaf plant image data in real time and store them in a trusted data block. When early disease and pest identification is performed, the system retrieves image data from the trusted data block, removes image noise points through preprocessing techniques such as median filtering, and unifies image resolution and color space to obtain standardized first image data.

[0078] Subsequently, the first image data is feature-extracted using a trained residual convolutional neural network (ResNet) to identify subtle features such as leaf spot morphology and color abnormalities, generating an image feature set. Meanwhile, reference feature sets are extracted from the early disease and pest feature library constructed in the base, and the dynamic time warping algorithm is used to calculate the similarity between the two. If the feature similarity value exceeds a preset threshold T1 (such as 0.75), it is determined that the three-leaf plant has early lesions, and the early disease and pest identification result is output.

[0079] After confirming the lesions, a random forest classification algorithm is used to determine the type of lesions, accurately distinguishing between disease types such as three-leaf root rot and black spot, and generating corresponding lesion type labels. Based on the label information, the occurrence rules and transmission conditions of the corresponding disease are retrieved from the disease knowledge base, such as three-leaf root rot being caused by excessive soil humidity and easily spreading in continuous cropping plots, thereby forming a disease distribution feature set.

[0080] Finally, a density peak clustering algorithm (DPC) is used to analyze the disease distribution, combined with data collected by the base's temperature and humidity sensor network, to find that three-leaf plants in low-lying and poorly drained areas have a significantly increased risk of root rot. From this, the high-risk disease distribution area is delineated, which can help staff accurately deploy biological control agent spraying and isolate monitoring of three-leaf plants in key areas, effectively reducing the risk of disease spread and ensuring the quality and yield stability of traditional Chinese medicine.

[0081] S2, according to the early identification result, fuse the temperature, humidity and other parameters in the environmental data, adopt a multi-modal data fusion algorithm, calculate the comprehensive disease risk index B (B = αT + βH + γI, where T is the temperature, H is the humidity, I is the image feature value, and α, β, γ are weight coefficients) through weighted average method to get the disease risk assessment result;

[0082] Further, in step S2, the disease risk assessment result is obtained, specifically including:

[0083] S21, obtain the temperature parameter and humidity parameter in the environmental data, then fuse the standardized environmental data set and the image feature data set, adopt a multi-modal data fusion algorithm to generate a fused feature data set;

[0084] S22, calculate the comprehensive disease risk index B by a weighted average method, obtain a risk index value, when the risk index value exceeds a preset threshold, determine that it is a high disease risk, and generate a high risk assessment result, when the risk index value is lower than the preset threshold, determine that it is a low disease risk, and generate a low risk assessment result;

[0085] wherein the comprehensive disease risk index B = aT + bH + gI, wherein T is temperature, H is humidity, I is image feature value, a, b, g are weight coefficients;

[0086] S23, generate corresponding disease risk assessment report data according to the high risk assessment result or the low risk assessment result.

[0087] In the intelligent monitoring process of the disease and pest of the standard base of traditional Chinese medicinal materials, after early identification, the disease risk assessment phase of step S2 is entered. The environmental data are collected in real time by the devices such as the temperature and humidity sensors and the soil moisture content monitors which are widely distributed in the base, the key parameters such as temperature, humidity and soil water content are accurately extracted from the data, the image feature data set which is standardized in the early stage is integrated, the multi-modal data fusion algorithm based on attention mechanism is used, the correlation between the environment and the image data is deeply mined, and the fusion feature data set containing the meteorological conditions, the soil conditions and the plant disease characteristics is generated.

[0088] Subsequently, the system calculates the comprehensive disease risk index B according to the formula B = aT + bH + gI. Wherein T is the real-time monitored temperature (such as 28℃), H is the humidity (75%), I is the disease degree value based on the image feature extraction (0.7), a, b, g are the weight coefficients calibrated by the historical disease and pest data and professional research (such as a = 0.25, b = 0.3, g = 0.45), and B = 0.25 x 28 + 0.3 x 75 + 0.45 x 0.7 = 30.85 is obtained by weighted calculation.

[0089] The calculation result is compared with the preset threshold 22, because 30.85 is higher than the threshold, it is determined that the area is in a high disease risk state, and a high risk assessment result is generated; otherwise, it is determined to be a low risk. For the high risk assessment result, the system automatically generates a detailed report, records key data such as soil humidity of 75%, leaf spot feature index of 0.7, analyzes the risk index calculation process, and gives precise prevention and control suggestions such as “immediately spray Bacillus subtilis biological agent on the diseased area, reduce the soil humidity to below 60%” according to the growth habit of Panax notoginseng; the low risk report focuses on recording data trend and prompting to continuously monitor the temperature and humidity changes of the key plots.

[0090] S3, according to the disease risk assessment results, using reinforcement learning algorithm, constructing state space S (including environmental data and disease risk index), action space A (including spraying pesticides, adjusting irrigation and other control measures) and reward function R (based on control effect and resource consumption), through Q-learning iterative updating strategy, obtaining dynamic optimized control strategy;

[0091] Further, in step S3, the dynamic optimized control strategy is obtained, specifically including:

[0092] S31, obtaining environmental data and disease risk assessment results, constructing state space S, containing temperature, humidity, soil parameters and risk index, determining the dimension of state space S, and then defining action space A through state space S, containing spraying pesticide types, dosage, adjusting irrigation frequency and water quantity, obtaining action space set;

[0093] S32, according to the control effect and resource consumption, constructing reward function R, when the control effect is higher than the preset threshold and the resource consumption is lower than the preset threshold, the reward value is updated positively, obtaining the reward function model;

[0094] S33, using Q-learning algorithm, initializing Q value table, updating Q value for state space S and action space A, obtaining dynamic strategy, and then extracting the optimal action combination of current state from the dynamic strategy, generating control scheme containing spraying pesticides and adjusting irrigation;

[0095] S34, obtaining environmental data and disease risk index after executing the control scheme, if the risk index is lower than the preset threshold, maintaining the current scheme, otherwise re-entering state space S to update, obtaining new control scheme;

[0096] S35, through multiple iterations, recording the control effect and resource consumption of each control scheme, updating reward function R, obtaining optimized dynamic control scheme.

[0097] When the disease and pest control work of the Chinese herbal medicine standard base is promoted to step S3, in order to formulate the dynamic optimized control strategy, the staff will comprehensively collect the environmental data of the base, including the temperature and humidity of the air, the pH value, fertility, air permeability of the soil and the light intensity and other information, and at the same time, combining with the disease risk assessment results in the early stage, constructing state space S, which will include temperature, humidity, soil parameters, light conditions and risk index and other key elements, and accurately determining its dimension.

[0098] Based on the state space S, the staff defines the action space A, which covers various prevention and control actions, such as selecting different types of biological agents, spraying chemical agents, determining the use dosage of agents, adjusting the frequency and water volume of irrigation, improving the ventilation conditions of the base, building shading facilities to adjust the light intensity, etc., thereby forming a set of action spaces.

[0099] Next, considering both the prevention and control effect and resource consumption, the reward function R is constructed. If the prevention and control effect of the disease and pest is higher than the preset threshold value after implementing the prevention and control measures, and the resource consumption (such as the use amount of pesticides, water resources, and electricity) is lower than the preset threshold value, the reward value will be updated positively, thereby obtaining the reward function model. Then, the Q-learning algorithm is used to initialize the Q value table. Based on the state space S and the action space A, the Q value is iteratively updated by continuously trying different actions and observing the corresponding results, and a dynamic strategy is gradually formed. From the dynamic strategy, the optimal action combination under the current base state is extracted to generate a prevention and control scheme including the selection of appropriate agent spraying, precise adjustment of irrigation, and light conditions.

[0100] S4, specific action instructions are extracted from the dynamically optimized prevention and control strategy, and the Internet of Things devices are used to control agricultural machinery to perform spraying agents or adjust irrigation, etc., to generate an execution log, which is stored in a blockchain account to obtain an implementation record of the prevention and control measures;

[0101] Further, in step S4, the implementation record of the prevention and control measures is obtained, which specifically includes:

[0102] Action instructions are extracted from the dynamically optimized prevention and control strategy, and natural language processing technology is used to analyze the strategy text to identify the instruction content of spraying agents or adjusting irrigation, thereby obtaining a set of action instructions;

[0103] When the set of action instructions includes spraying agent instructions, control signals are sent to agricultural machinery through Internet of Things devices to obtain the running state of the machinery and determine the execution of the spraying agent task. When the set of action instructions includes irrigation adjustment instructions, control signals are sent to agricultural machinery through Internet of Things devices to obtain irrigation parameter adjustment data and determine the execution of the irrigation task;

[0104] According to the spraying agent task performed by the agricultural machinery, a log generation algorithm is used to record the agent type, spraying time, and area information to obtain a spraying execution log. According to the adjustment irrigation task performed by the agricultural machinery, a log generation algorithm is used to record the irrigation time, water volume, and area information to obtain an irrigation execution log;

[0105] The spraying execution log and the irrigation execution log are uploaded to the blockchain ledger through a blockchain ledger interface, a storage transaction hash value is obtained, log storage is determined to be complete, the stored execution log is extracted from the blockchain ledger, a data integration technology is used to generate a prevention and control measure implementation record containing a timestamp, a task type and a region, and a final implementation record is obtained.

[0106] Specifically, the action instructions in the dynamically optimized prevention and control strategy are accurately parsed through natural language processing technology to quickly generate an action instruction set, ensuring that the prevention and control instructions are accurately conveyed. Remote control and real-time state monitoring of agricultural machinery are achieved with the help of Internet of Things devices. Whether it is pesticide spraying or irrigation adjustment tasks, they can be accurately executed and execution data can be obtained to ensure that the prevention and control measures are efficiently implemented. The log generation algorithm comprehensively records the agricultural machinery operation information to realize digital storage of the prevention and control process. Based on the blockchain technology, the execution log is uploaded. The characteristics of non-tamperability and traceability ensure the security and credibility of the data. Finally, the data integration technology is used to generate the final implementation record to form a complete and transparent prevention and control measure implementation file. This not only provides a reliable basis for subsequent prevention and control effect evaluation, but also helps management personnel to review and optimize the prevention and control process, improving the intelligent and fine level of disease and pest control in traditional Chinese medicinal material standard bases, and realizing the whole-process closed-loop management from instruction issuance to execution record.

[0107] S5, obtaining the implementation record and real-time environmental data, using a time series analysis algorithm to calculate the change trend of the disease risk index, if the change trend shows that the risk index decreases and is lower than a preset threshold T2 (T2 = 0.3), it is determined that the prevention and control strategy is effective, and a prevention and control effect evaluation result is obtained;

[0108] Further, in step S5, the prevention and control effect evaluation result is obtained, specifically including:

[0109] Obtaining the implementation record data and real-time environmental data, generating a standardized time series data set through data cleaning and formatting processing, and then using a time series analysis algorithm to process the time series data set to calculate the disease risk index and obtain a risk index sequence;

[0110] The risk index sequence is analyzed by a sliding window method to calculate the change trend of the risk index, and a trend feature sequence is obtained. When the trend feature sequence shows that the risk index continues to decrease, a comparison method is used to compare with a preset threshold T2 to determine whether it is lower than the threshold, and a threshold comparison result is obtained;

[0111] According to the threshold comparison result, a logical decision rule is used. When the risk index is lower than the preset threshold T2, it is determined that the prevention and control strategy is effective, and a strategy effectiveness identifier is obtained;

[0112] The prevention and treatment effect evaluation data is generated by aggregating the strategy effectiveness identifier and the risk index sequence, and the evaluation result dataset is obtained. The dynamic display data of the prevention and treatment effect is obtained by processing the evaluation result dataset using a visualization technology to generate a change trend chart.

[0113] In the pest control work, the prevention and treatment effect evaluation link of step S5 plays a key role. By comprehensively collecting implementation records and real-time environmental data, standardizing processing and in-depth analysis, the disease risk index and its change trend can be accurately calculated. When the risk index continues to decline and is lower than the preset threshold, it indicates that the control strategy is effective. After the effectiveness identifier and the risk index sequence are aggregated and visualized, the change trend chart can intuitively present the prevention and treatment effect. The staff can clearly master the development trend of pests and diseases and intuitively evaluate the effectiveness of the control strategy, thereby providing a key reference for subsequent adjustment of control measures and reasonable allocation of resources, effectively ensuring the healthy production of Chinese herbal medicine standard bases and reducing economic losses caused by pests and diseases.

[0114] S6, updating the reward function of the reinforcement learning algorithm according to the prevention and treatment effect evaluation result, adjusting the reward weight to improve the preference for low resource consumption strategies, generating an updated dynamic prevention and treatment strategy, storing in a distributed database, and obtaining adaptive prevention and treatment strategy optimization results;

[0115] Further, in step S6, the adaptive prevention and treatment strategy optimization results are obtained, specifically including:

[0116] Key indicator data is obtained from the prevention and treatment effect evaluation result, and statistical analysis method is used to calculate the effect score and resource consumption score to obtain an evaluation index set. When the effect score is lower than the preset threshold, the reward function of the reinforcement learning model is optimized by the gradient descent algorithm to adjust the reward weight, and the updated reward function is obtained;

[0117] According to the updated reward function, a dynamic prevention and treatment strategy is generated using a reinforcement learning algorithm, the adaptive parameter set of the strategy is determined, and resource consumption features are extracted from the dynamic prevention and treatment strategy to determine whether the low resource consumption condition is met. When it is met, it is stored in a distributed database to obtain a candidate strategy set.

[0118] The candidate strategy set is verified in parallel using a distributed computing framework to obtain the performance indicators of the verified strategy, determine the preferred strategy set, and update the initial parameters of the reinforcement learning model through the preferred strategy set to generate the final adaptive prevention and treatment strategy, which is stored in the distributed database to obtain the optimized strategy result;

[0119] The optimized strategy result is extracted from the distributed database, and a consistency checking algorithm is used to verify the data integrity to obtain the final prevention and treatment strategy output.

[0120] Specifically, by deeply mining and analyzing the prevention and control effect evaluation results, the problems and deficiencies of the prevention and control strategy can be accurately identified. After obtaining the key indicator data and calculating the effect score and resource consumption score, the performance of the strategy in prevention and control effect and resource utilization can be determined. If the effect score is not good, the reward function of the reinforcement learning model is optimized by gradient descent algorithm, the reward weight is adjusted, the model is more suitable for actual needs, and the dynamic prevention and control strategy generated based on the updated reward function is screened through resource consumption judgment, distributed parallel verification and other links, and the optimal strategy considering efficient prevention and control and low resource consumption is selected, the initial parameters of the reinforcement learning model are updated, and the final adaptive prevention and control strategy is generated. This process not only effectively improves the accuracy and effectiveness of the prevention and control strategy, but also realizes the optimization of resource allocation. Through consistency check to ensure data reliability, it provides scientific, dynamic and efficient decision-making basis for disease and pest control in Chinese herbal medicine standard base, greatly enhances the ability of Chinese herbal medicine standard base to respond to diseases and pests, and reduces the long-term prevention and control cost.

[0121] S7, extract the latest strategy from the adaptive prevention and control strategy optimization result, combine the real-time data processing module, and generate continuously updated disease and pest control instructions by cyclically executing the closed-loop process of data acquisition, risk assessment and strategy optimization, store them in the blockchain ledger, and obtain the running result of the whole-process intelligent prevention and control system.

[0122] Further, in step S7, the running result of the whole-process intelligent prevention and control system is obtained, specifically including:

[0123] Obtain the latest adaptive strategy from the blockchain ledger, extract the strategy parameters using the data analysis module, and obtain the strategy parameter set;

[0124] Obtain environmental data through the real-time data acquisition module, clean and standardize the data using the preprocessing algorithm, and obtain the standardized data set. When the key indicators in the standardized data set exceed the preset threshold, calculate the disease and pest risk probability using the risk assessment model, and obtain the risk assessment result;

[0125] According to the risk assessment result, optimize the parameters of the adaptive strategy using the reinforcement learning algorithm, generate a dynamically adjusted prevention and control instruction set, store the prevention and control instruction set to the blockchain ledger through the instruction distribution module, verify the data integrity using the consistency check algorithm, and obtain the verified instruction set;

[0126] Extract the execution parameters from the verified instruction set, update the sampling frequency of the real-time data acquisition module using the closed-loop feedback mechanism, obtain the optimized running result, and perform parallel analysis of the execution effect of the prevention and control instructions using the distributed computing framework, and obtain the updated adaptive strategy parameter set.

[0127] Specifically, the parameters of the latest adaptive strategy are extracted from the blockchain ledger, and the environmental data is collected and preprocessed in real time. If the key indicators exceed the threshold, the risk assessment model is used to calculate the risk probability of pests and diseases. According to the evaluation results, the strategy parameters are optimized through reinforcement learning algorithm to generate prevention and control instruction set, which is stored in the blockchain ledger after consistency verification. Then, the execution parameters are extracted, and the real-time data collection frequency is adjusted using the closed-loop feedback mechanism. The instruction execution effect is analyzed by using the distributed computing framework, and finally the adaptive strategy parameters are updated to obtain the operation results of the intelligent prevention and control system throughout the whole process, realizing the accurate, efficient and dynamic prevention and control of pests and diseases in Chinese herbal medicine standard bases.

[0128] Further, before extracting the plant image data from the trusted data block, it further includes:

[0129] Through the sensor network and the Internet of Things device, temperature, humidity, soil nutrients, light intensity and other environmental data are obtained from the Chinese herbal medicine planting environment. Combined with the plant image data captured by the camera, a multi-modal data set containing a time stamp is generated and stored in a distributed database to obtain a data set of environmental and plant states. Through blockchain technology, the data set is encrypted, and the data source and integrity are verified through a smart contract to generate a data block containing a hash value and store it in a distributed ledger to obtain a trusted data transmission and storage result.

[0130] Further, the data set of environmental and plant states specifically includes:

[0131] Temperature data, humidity data, soil nutrients, light intensity collected by the sensor network and Internet of Things device, and plant images collected by the camera are generated into a multi-modal data set with a time stamp and stored in a distributed database. Then, the time stamp is queried through the distributed database to extract the environmental state and plant state in the multi-modal data set. When the temperature data, humidity data, soil nutrients or light intensity in the environmental state exceed the preset threshold, it is marked as abnormal environmental data to obtain an abnormal environmental data set.

[0132] The plant image is processed using a convolutional neural network algorithm to extract the corresponding plant state from the abnormal environmental data set. If the leaf color or shape in the plant image deviates from the preset health standard, it is marked as abnormal plant data to obtain an abnormal plant data set. According to the abnormal environmental data set and the abnormal plant data set, the time series correlation between the environmental state and the plant state is calculated, and the correlation is calculated using the Pearson correlation coefficient to obtain a correlation coefficient set of the environment and the plant;

[0133] Through the correlation coefficient set, the environmental state parameters and plant state parameters with absolute values higher than the preset threshold are extracted. If the environmental state parameters in the combination are temperature data or humidity data, an environmental regulation instruction is generated to obtain a regulation instruction set.

[0134] According to the regulation instruction set, a control signal for the Internet of Things device is generated, the collection value of the temperature data or the humidity data is adjusted, the adjusted multi-modal data set is stored to the distributed database, the optimized data set is obtained, the plant state is extracted from the optimized data set, the random forest algorithm is used to predict the plant growth trend, and the plant growth prediction result is obtained.

[0135] Specifically, a closed-loop intelligent monitoring and regulation mechanism is embodied, and through real-time monitoring, data analysis, anomaly identification, environment regulation and growth prediction, fine management and intelligent regulation of the Chinese herbal medicine planting environment are realized.

[0136] Further, a trusted data transmission and storage result is obtained, specifically including:

[0137] The initial data set is encrypted by the block chain technology to generate an encrypted data set, if the encrypted data set passes the preset encryption strength verification, the data source and integrity of the encrypted data set are verified by the smart contract to obtain a verification result, and according to the verification result, a data block containing a hash value is generated;

[0138] The data block is stored in the distributed ledger, a storage address is obtained, the data block in the storage address is verified by the hash value, the data integrity is judged, if the data integrity passes the verification, a trusted data transmission record is obtained, and according to the transmission record, a trusted data storage result is generated.

[0139] Embodiment 2

[0140] Please refer to Figure 4 The embodiment provides a Chinese herbal medicine standard base pest and disease intelligent monitoring and precise prevention and treatment system, which is used for realizing a Chinese herbal medicine standard base pest and disease intelligent monitoring and precise prevention and treatment method, and comprises:

[0141] The data acquisition and verification module collects key environmental data and plant images in the Chinese herbal medicine planting environment by using a sensor network and an Internet of Things device, creates a multi-modal data set with a time stamp and stores it in a distributed database, uses a block chain technology for encryption and integrity verification, generates a data block containing a hash value, ensures the safety and credibility of the data, and stores it in a distributed ledger;

[0142] The early pest and disease identification module extracts features from the plant images by using a convolutional neural network model, compares them with an early pest and disease feature library, identifies early lesions, distinguishes lesion types by using a classification algorithm, and obtains corresponding disease distribution characteristics from a disease knowledge base to analyze the disease distribution area;

[0143] The disease risk assessment module fuses environmental data and image feature data, adopts a multi-modal data fusion algorithm to calculate a comprehensive disease risk index, generates a disease risk assessment report according to the risk assessment result, and provides a basis for prevention and control decision-making.

[0144] The prevention and control strategy generation module adopts a reinforcement learning algorithm to construct a state space, an action space and a reward function, iteratively updates the strategy through Q-learning, and generates a dynamically optimized prevention and control strategy to achieve precise prevention and control.

[0145] The prevention and control measure implementation and recording module extracts specific action instructions from the dynamically optimized prevention and control strategy, controls agricultural machinery to perform spraying or irrigation adjustment through Internet of Things devices, generates an execution log, and stores it in a blockchain ledger to record the implementation of the prevention and control measures.

[0146] The effect evaluation and strategy optimization module obtains the implementation records and real-time environmental data, adopts a time series analysis algorithm to calculate the change trend of the disease risk index, updates the reward function of the reinforcement learning algorithm according to the evaluation result, generates adaptive prevention and control strategy optimization results, and realizes continuous optimization of the prevention and control strategy.

[0147] The intelligent prevention and control system operation module extracts the latest strategy from the adaptive prevention and control strategy optimization results, combines the real-time data processing module, and through the closed-loop process of cyclically executing data collection, risk assessment and strategy optimization, generates continuously updated disease and pest control instructions, stores them in a blockchain ledger, and realizes the operation of the whole-process intelligent prevention and control system.

[0148] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to make equivalent embodiments with equivalent changes, without departing from the scope of the technical solution of the present application. Any modification, equivalent change and modification of the above embodiments made in accordance with the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. Intelligent monitoring and precise control method for pests and diseases in standard bases for Chinese medicinal materials, characterized by: The steps include: Extract plant image data from the trusted data block and use a convolutional neural network model to extract features from the image. Combined with a pre-established early feature library of pests and diseases, when the similarity between the extracted features and the features in the library exceeds a preset threshold T1, it is judged as an early lesion and the early identification result of pests and diseases is obtained. Based on the early identification results, the parameters in the environmental data are integrated, and a multimodal data fusion algorithm is used to calculate the comprehensive disease risk index through the weighted average method to obtain the disease risk assessment result; Based on the results of disease risk assessment, a reinforcement learning algorithm is used to construct the state space, action space, and reward function. Through Q-learning iterative update strategy, a dynamically optimized prevention and control strategy is obtained. Extract specific action instructions from the dynamically optimized prevention and control strategy, control agricultural machinery through IoT devices to spray pesticides or adjust irrigation operations, generate execution logs, store them in the blockchain ledger, and obtain a record of the implementation of prevention and control measures; Obtain implementation records and real-time environmental data, use time series analysis algorithms to calculate the changing trend of the disease risk index. If the changing trend shows that the risk index is decreasing and is lower than the preset threshold T2, the prevention and control strategy is judged to be effective, and the prevention and control effect evaluation results are obtained; Based on the results of the control effect evaluation, the reward function of the reinforcement learning algorithm is updated. By adjusting the reward weight, the preference for low resource consumption strategies is increased. An updated dynamic control strategy is generated and stored in a distributed database to obtain the adaptive control strategy optimization results. The latest strategy is extracted from the optimization results of the adaptive prevention and control strategy. Combined with the real-time data processing module, a closed-loop process of data collection, risk assessment and strategy optimization is cyclically executed to generate continuously updated pest and disease control instructions, which are stored in the blockchain ledger to obtain the operation results of the full-process intelligent prevention and control system.

2. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: Obtain early identification results of pests and diseases, including: Acquire plant image data from the trusted data block, use preprocessing technology to denoise and standardize the image to obtain first image data, and then use a convolutional neural network to extract features from the first image data to generate an image feature set; Obtain a reference feature set from a pre-established early feature library of pests and diseases, calculate the similarity between the image feature set and the reference feature set, and obtain a feature similarity value. When the feature similarity value exceeds a preset threshold T1, it is judged as an early lesion and an early identification result of pests and diseases is generated; Based on the early identification results of pests and diseases, a classification algorithm is used to distinguish the lesion types and obtain lesion type labels. Through the lesion type labels, the corresponding disease distribution characteristics are obtained from the pre-established disease knowledge base to generate a disease distribution feature set; According to the disease distribution feature set, clustering algorithm is used to perform partition analysis on disease distribution and obtain the disease distribution area results.

3. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: Obtain disease risk assessment results, including: Obtain temperature and humidity parameters from environmental data, fuse the standardized environmental dataset and image feature dataset, and use a multimodal data fusion algorithm to generate a fused feature dataset. The comprehensive disease risk index is calculated by the weighted average method to obtain a risk index value. When the risk index value exceeds the preset threshold, it is determined to be a high disease risk and a high risk assessment result is generated. When the risk index value is lower than the preset threshold, it is determined to be a low disease risk and a low risk assessment result is generated. Based on the high-risk assessment results or low-risk assessment results, corresponding disease risk assessment report data is generated.

4. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: The dynamically optimized prevention and control strategies include: Obtain environmental data and disease risk assessment results, construct a state space, including temperature, humidity, soil parameters and risk index, determine the state space dimension, and then define the action space based on the state space, including the type and dosage of spraying pesticides, and adjust the irrigation frequency and water volume, to obtain the action space set; Based on the prevention and control effect and resource consumption, a reward function is constructed. When the prevention and control effect is higher than the preset threshold and the resource consumption is lower than the preset threshold, the reward value is positively updated to obtain the reward function model; Using the Q-learning algorithm, the Q-value table is initialized. The Q-value is iteratively updated for the state space and action space to obtain a dynamic strategy. The optimal action combination for the current state is then extracted from the dynamic strategy to generate a prevention and control plan that includes spraying pesticides and adjusting irrigation. Obtain environmental data and disease risk index after the prevention and control plan is implemented. If the risk index is lower than the preset threshold, maintain the current plan; otherwise, re-enter the state space update to obtain a new prevention and control plan. Through multiple iterations, the control effect and resource consumption of each control plan are recorded, the reward function is updated, and the optimized dynamic control plan is obtained.

5. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: Obtain implementation records of prevention and control measures, including: Extract action instructions from the dynamic optimization control strategy, use natural language processing technology to parse the strategy text, identify the instructions for spraying pesticides or adjusting irrigation, and obtain the action instruction set; When the action instruction set includes a spraying instruction, a control signal is sent to the agricultural machinery through the IoT device to obtain the machine's operating status and determine the execution of the spraying task. When the action instruction set includes an irrigation adjustment instruction, a control signal is sent to the agricultural machinery through the IoT device to obtain irrigation parameter adjustment data and determine the execution of the irrigation task. According to the spraying task executed by agricultural machinery, a log generation algorithm is used to record the type of pesticide, spraying time and area information to obtain a spraying execution log. Then, according to the adjustment irrigation task executed by agricultural machinery, a log generation algorithm is used to record the irrigation time, water volume and area information to obtain an irrigation execution log. Through the blockchain ledger interface, the spraying execution log and irrigation execution log are uploaded to the blockchain ledger, the stored transaction hash value is obtained, and the log storage is confirmed to be completed. Then, the stored execution log is extracted from the blockchain ledger, and data integration technology is used to generate the prevention and control measures implementation record containing timestamp, task type and area to obtain the final implementation record.

6. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: The results of the prevention and control effect evaluation are obtained, including: Obtain implementation record data and real-time environmental data, generate a standardized time series data set through data cleaning and formatting, then use the time series analysis algorithm to process the time series data set, calculate the disease risk index, and obtain the risk index sequence; The risk index sequence is analyzed by the sliding window method, and the changing trend of the risk index is calculated to obtain the trend feature sequence. When the trend feature sequence shows that the risk index continues to decline, it is compared with the preset threshold T2 through the comparison method to determine whether it is lower than the threshold and obtain the threshold comparison result; According to the threshold comparison results, a logical judgment rule is adopted. When the risk index is lower than the preset threshold T2, the prevention and control strategy is judged to be effective and the strategy effectiveness mark is obtained; By summarizing the strategy effectiveness identification and risk index sequence, we generate prevention and control effect evaluation data and obtain the evaluation result data set. We then use visualization technology to process the evaluation result data set, generate a change trend chart, and obtain dynamic display data of the prevention and control effect.

7. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: The adaptive prevention and control strategy optimization results are obtained, including: Key indicator data is obtained from the control effect evaluation results. Statistical analysis methods are used to calculate the effect score and resource consumption score to obtain the evaluation indicator set. When the effect score is lower than the preset threshold, the reward function of the reinforcement learning model is optimized through the gradient descent algorithm, and the reward weight is adjusted to obtain the updated reward function. Based on the updated reward function, a reinforcement learning algorithm is used to generate a dynamic prevention and control strategy, determine the strategy's adaptive parameter set, and then extract resource consumption characteristics from the dynamic prevention and control strategy to determine whether the low resource consumption condition is met. If so, the characteristics are stored in a distributed database to obtain a candidate strategy set. A distributed computing framework is used to perform parallel verification of candidate strategy sets, obtain the verified strategy performance indicators, determine the strategy set, and then use the strategy set to update the initial parameters of the reinforcement learning model to generate the final adaptive prevention and control strategy, which is stored in a distributed database to obtain the optimized strategy results. The optimized strategy results are extracted from the distributed database, and the data integrity is verified using a consistency check algorithm to obtain the final prevention and control strategy output.

8. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 1, characterized in that: Before extracting the plant image data from the trusted data block, it also includes: Through sensor networks and IoT devices, environmental data such as temperature, humidity, soil nutrients, and light intensity are obtained from the Chinese medicinal materials planting environment. Combined with the plant image data captured by the camera, a multimodal data set containing timestamps is generated and stored in a distributed database to obtain a data set of the environment and plant status. The data set is encrypted through blockchain technology, and the data source and integrity are verified through smart contracts. Data blocks containing hash values ​​are generated and stored in a distributed ledger to obtain reliable data transmission and storage results.

9. The method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to claim 8, characterized in that: Obtain reliable data transmission and storage results, including: The initial data set is encrypted using blockchain technology to generate an encrypted data set. If the encrypted data set passes the preset encryption strength verification, the smart contract is used to verify its data source and integrity, and the verification result is obtained. Based on the verification result, a data block containing a hash value is generated; The data blocks are stored in the distributed ledger, the storage address is obtained, and the hash value is used to verify the data blocks in the storage address to determine the data integrity. If the data integrity passes the verification, the trusted data transmission record is obtained, and the trusted data storage result is generated based on the transmission record.

10. A system for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials, applied to the method for intelligent monitoring and precise control of pests and diseases in a standard base of Chinese medicinal materials according to any one of claims 1 to 9, characterized in that: include: The data collection and verification module uses sensor networks and IoT devices to collect key environmental data and plant images from the TCM planting environment, creates a multimodal dataset with timestamps, and stores it in a distributed database. Blockchain technology is used for encryption and integrity verification, generating data blocks containing hash values. The early identification module for pests and diseases uses a convolutional neural network model to extract features from plant images and compares them with a database of early pest and disease features to identify early lesions. It then uses a classification algorithm to distinguish lesion types and obtains corresponding disease distribution features from a disease knowledge base to analyze disease distribution areas. The disease risk assessment module integrates environmental data and image feature data, uses multimodal data fusion algorithm to calculate the comprehensive disease risk index, and generates a disease risk assessment report based on the risk assessment results. The prevention and control strategy generation module uses reinforcement learning algorithms to construct state space, action space, and reward function, and iteratively updates the strategy through Q-learning to generate dynamically optimized prevention and control strategies; The prevention and control measures implementation and recording module extracts specific action instructions from the dynamically optimized prevention and control strategies, controls agricultural machinery through IoT devices to perform operations such as spraying pesticides or adjusting irrigation, generates execution logs, stores them in the blockchain ledger, and records the implementation of prevention and control measures; The effect evaluation and strategy optimization module obtains implementation records and real-time environmental data, uses a time series analysis algorithm to calculate the changing trend of the disease risk index, updates the reward function of the reinforcement learning algorithm based on the evaluation results, and generates adaptive control strategy optimization results; The intelligent pest control system operation module extracts the latest strategy from the adaptive prevention and control strategy optimization results, combines it with the real-time data processing module, and generates continuously updated pest control instructions through a closed-loop process of cyclically executing data collection, risk assessment and strategy optimization, which are stored in the blockchain ledger.

Images