Monitoring and control method and system based on the occurrence law of pests and diseases of pitaya

Abstract

The present invention discloses a monitoring and prevention method and system based on the occurrence law of pitaya pests, including: retrieving and obtaining historical pest occurrence data of pitaya for data preprocessing, dividing and generating a subset of pest occurrence data with pest category labels; analyzing the occurrence causal characteristics corresponding to the pest data sample and the causal relationship between the characteristics in the pest occurrence data subset, and constructing a pest attribution path; obtaining the similarity between the multi-source monitoring data and the pest attribution path in the monitoring area, and generating a pitaya pest warning in the monitoring area through the similarity; extracting the pest category and predicted hazard status corresponding to the pitaya pest warning, and analyzing and obtaining a combination of prevention and control methods. The present invention provides an accurate monitoring and early warning model for pitaya pests, and formulates a scientific prevention and control plan in a targeted manner according to the predicted hazard status, improves the prevention and control effect of pitaya pests, and effectively improves the quality of pitaya.

Description

Technical Field

[0001] The present invention relates to the technical field of pest control, and more specifically, to a monitoring and control method and system based on the occurrence law of dragon fruit pests. Background Art

[0002] Dragon fruit is rich in nutrients and has high nutritional value. There are many challenges in the cultivation process of dragon fruit, especially since it needs to grow in high temperature and high humidity areas. It is very susceptible to diseases and insect pests, which cause a variety of diseases and insect pests. These diseases and insect pests seriously affect the growth of dragon fruit plants, fruit quality and economic benefits. Harmful organisms include pathogens and pests that cause diseases and insect pests, as well as mollusks such as snails and slugs. In order to reduce the impact of diseases and insect pests on the quality and yield of dragon fruit, corresponding measures must be taken to prevent and control diseases and insect pests. Although a variety of prevention and control measures have been used to prevent and control dragon fruit diseases and insect pests, there are still certain limitations.

[0003] At present, agricultural production is still in the stage of using a large number of fertilizers and pesticides, mainly relying on traditional chemical control methods to control crop diseases and insect pests. The abuse and misuse of pesticides have brought about increasingly serious problems such as reduced biodiversity, increased pest resistance, pesticide residues and environmental pollution. In addition, the traditional prediction of dragon fruit pests is mainly based on the experience of growers, who judge the possibility and type of pests and diseases, and then formulate corresponding prevention and control measures. This method of relying on experience judgment is highly subjective, has a low accuracy rate, and is prone to waste and loss of human and material resources. Therefore, how to accurately predict the occurrence trend of dragon fruit pests and formulate personalized prevention and control plans to improve the efficiency of pest control is an urgent problem to be solved. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a monitoring and prevention method and system based on the occurrence laws of pitaya pests, with the aim of analyzing the occurrence laws of the main diseases and pests of pitaya pests, accurately preventing and controlling pitaya pests, and providing a reference for pitaya pest prevention and control.

[0005] The first aspect of the present invention provides a monitoring and control method based on the occurrence law of dragon fruit pests, comprising:

[0006] Retrieving and obtaining historical pest occurrence data of pitaya for data preprocessing, dividing the historical pest occurrence data according to pest categories, and generating a subset of pest occurrence data with pest category labels;

[0007] Extracting features from the pest occurrence data subset, analyzing occurrence causal features corresponding to the pest data sample and the causal relationship between the features, and constructing a pest attribution path based on the occurrence causal features and the causal relationship;

[0008] Obtain multi-source monitoring data within the monitoring area, construct a multi-source monitoring data sequence based on the multi-source monitoring data, calculate the similarity between the multi-source monitoring data sequence and the pest attribution path, and generate a warning for pitaya pests in the monitoring area through the similarity.

[0009] Extract the pest categories and predicted damage status corresponding to the pitaya pest warning, and obtain a combination of control methods according to the pest categories and predicted damage status.

[0010] In this solution, the historical pest occurrence data is divided according to pest categories to generate subsets of pest occurrence data with pest category labels, specifically:

[0011] Obtain the historical pest occurrence data after data cleaning and normalization, use keyword extraction to obtain pest category keywords in the historical pest occurrence data, and calculate the semantic similarity between different pest category keywords;

[0012] Refine the extracted pest category keywords through the semantic similarity, merge the pest category keywords with the same semantic reference, and generate pest category labels after redundancy removal processing;

[0013] Use the pest category labels for clustering to obtain subsets of pest occurrence data under each pest category label, count the number of pest data samples included in different subsets of pest occurrence data, and screen out subsets of pest occurrence data where the number of pest data samples does not meet the preset quantity standard;

[0014] Perform upsampling on the selected subsets of pest occurrence data for data balancing processing, and output subsets of pest occurrence data with pest category labels after the sample quantity meets the standard.

[0015] In this solution, feature extraction is performed on the subsets of pest occurrence data to analyze the occurrence cause characteristics corresponding to pest data samples, specifically:

[0016] Obtain pest occurrence-related data in each subset of pest occurrence data to generate pest data samples, read the data change amplitude according to the pest data samples, initially select the influencing factors of different pitaya pests through the data change amplitude, and perform variance filtering on the initially selected influencing factors to generate an influencing factor set;

[0017] Feature extraction is performed on the set of influencing factors through a genetic algorithm. The set of influencing factors is initialized to generate an initial population, and the initial population is encoded. A pest prediction model is trained using the method of cross-validation, and the average accuracy is calculated as the fitness value of an individual.

[0018] Based on the fitness value, a preset number of individuals are selected from the population according to the elite selection method, and the direction of crossover and mutation is generated by introducing differential calculation. The population is updated through mutation and differential operations. When the preset minimum fitness threshold is reached, the subset of influencing factors is output.

[0019] On the basis of the subset of influencing factors, the pest prediction model is retrained. The mean square error of each influencing factor is obtained by using the SVM training model as the influencing factor score. The influencing factors are further screened according to the influencing factor score to obtain a new subset of influencing factors.

[0020] The pest prediction model is trained again using the new subset of influencing factors until the model obtains the optimal value, and then the final subset of influencing factors is output to obtain the occurrence causative features corresponding to the corresponding pest category label.

[0021] In this solution, the causal relationship between the occurrence causative features is obtained as follows:

[0022] The occurrence causative features corresponding to different pest categories are obtained. Based on the correlation between the occurrence causative features, the information flow path between the occurrence causative features is initially defined, and the transfer entropy between the occurrence causative feature nodes in different information flow paths is calculated.

[0023] The total transfer entropy between the occurrence causative feature nodes in the information flow path is obtained, and the information flow path with the largest total transfer entropy is selected. The causal relationship between the occurrence causative features during the pest occurrence process is characterized according to the selected information flow path.

[0024] In this solution, a pest attribution path is constructed based on the occurrence causative features and the causal relationship as follows:

[0025] According to the historical pest occurrence data, the critical value interval of the occurrence causative features corresponding to the occurrence of dragon fruit pests is obtained, and the pest occurrence probability when the parameters corresponding to the occurrence causative features are within the critical value interval is judged based on the historical dragon fruit planting data.

[0026] The weight information of each occurrence causative feature is generated according to the pest occurrence probability, the information flow path corresponding to the causal relationship is obtained, the weight information is matched with the occurrence causative feature nodes, and the critical value interval is marked for the causal relationship pointing to represent the attribute information of the causal relationship pointing.

[0027] Generate a pest attribution path using the information flow path after marker matching.

[0028] In this solution, calculate the similarity between the multi-source monitoring data sequence and the pest attribution path, and generate a warning for pitaya pests in the monitoring area based on the similarity. Specifically:

[0029] Obtain the occurrence cause characteristics corresponding to different pest categories, aggregate and de-duplicate all the occurrence cause characteristics to generate the monitoring data categories for the pitaya planting area, obtain multi-source monitoring data based on the monitoring data categories, and preprocess and combine with time series to construct a multi-source monitoring data sequence;

[0030] Based on the pest attribution path, obtain the occurrence cause characteristic nodes that fall into the critical value interval according to the multi-source monitoring data sequence, and obtain the highly associated occurrence cause characteristic nodes corresponding to the occurrence cause characteristic nodes that fall into the critical value interval according to the connection relationship of the pest attribution path;

[0031] Perform path fitting according to the occurrence cause characteristic nodes that fall into the critical value interval and the highly associated occurrence cause characteristic nodes, segment the fitted path, and within each segment, use the Fréchet distance after node weighting as the basis for similarity quantification to calculate the average similarity between the fitted path and the pest attribution path;

[0032] Obtain the pest categories corresponding to the pest attribution paths with an average similarity greater than the preset similarity threshold, and obtain the damage status according to the pest occurrence symptoms of the pitaya plants. Generate a warning for pitaya pests based on the pest categories and the damage status.

[0033] In this solution, obtain a combination of control methods according to the pest categories and the predicted damage status. Specifically:

[0034] Read the corresponding pest categories and damage status through the pitaya pest warning, extract the change characteristics of the monitoring data according to the multi-source monitoring data sequence, train the mapping relationship between the multi-source monitoring data and the pest damage status using historical pest occurrence data, and obtain the predicted damage status after a preset time according to the mapping relationship;

[0035] Judge whether the predicted damage status exceeds the preset damage status range. When it does not exceed, select the green control measures for the pest category through the relevant knowledge graph, retrieve the control examples of each green control measure, obtain the average control time based on the control examples, and generate a ranking of green control measures;

[0036] Evaluate the control time through the predicted damage status as the control time standard, and select a preset number of green control measures that meet the control time standard according to the ranking of the green control measures for combination to generate a combination of control methods;

[0037] When it exceeds, recommended pesticides for pest categories are selected through the relevant knowledge graph, and the corresponding relationship between different pesticide application concentrations and the average control time is obtained according to the predicted damage situation;

[0038] The longest control time is obtained according to the control time standard, the pesticide application concentration for the longest control time is obtained through the above corresponding relationship, the recommended pesticide and the pesticide application concentration are used to generate chemical control measures, and they are combined with the green control measures with the shortest average control time to generate a combined control method.

[0039] The second aspect of the present invention provides a monitoring and control system based on the occurrence law of pests on pitaya, including a pest occurrence law analysis module, a multi-source monitoring data acquisition module, a pitaya pest warning module, a control decision-making module and a data visualization module;

[0040] The pest occurrence law analysis module is responsible for analyzing the occurrence cause characteristics corresponding to different pest categories and the causal relationship between the characteristics according to the historical pest occurrence data, constructing the corresponding pest attribution path, and representing the occurrence law of pitaya pests;

[0041] The multi-source monitoring data acquisition module is responsible for collecting multi-source monitoring data in the monitoring area, performing serialization processing, and constructing a multi-source monitoring data sequence with the same structure as the pest attribution path;

[0042] The pitaya pest warning module is responsible for calculating the similarity between the multi-source monitoring data sequence and different pest attribution paths, and generating a pitaya pest warning according to the similarity judgment;

[0043] The control decision-making module is responsible for obtaining a combined control method dominated by green control measures based on the warned pest category and the predicted damage situation;

[0044] The data visualization module is responsible for displaying the combined control method in the monitoring area, and performing real-time visualization display on the multi-source monitoring data and the pitaya pest warning.

[0045] Compared with the prior art, the beneficial effects of the present disclosure are:

[0046] The present invention provides an accurate monitoring and warning model for the damage of pests on pitaya, scientifically guides the prevention and control of pests, and specifically formulates a scientific control plan according to the predicted damage situation, improving the prevention and control effect of pitaya pests.

[0047] Formulate a scientific prevention and control plan with green prevention and control technology as the leading factor, enhance the ability of pitaya growers to emergently prevent and control harmful organisms, and also reduce the pollution caused by pesticides and their waste to the orchard. Gradually change the extensive management planting method of growers by increasing cultivation management improvement measures, and effectively improve the quality of pitaya. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments or exemplifications of the present invention, the following will briefly introduce the drawings required for use in the description of the embodiments or exemplifications. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the ones shown in these drawings.

[0049] Figure 1 Shows the flowchart of the monitoring and control method based on the occurrence law of harmful organisms of pitaya of the present invention;

[0050] Figure 2 Shows the flowchart of obtaining the cause characteristics of occurrence in the embodiment of the present invention;

[0051] Figure 3 Shows the flowchart of generating the early warning of harmful organisms of pitaya in the monitoring area in the embodiment of the present invention;

[0052] Figure 4 Shows the block diagram of the monitoring and control system based on the occurrence law of harmful organisms of pitaya of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] In order to be able to more clearly understand the above-mentioned objects, features and advantages of the present invention, the following will further describe the present invention in detail in combination with the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0054] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0055] Figure 1 Shows the flowchart of the monitoring and control method based on the occurrence law of harmful organisms of pitaya of the present invention.

[0056] As Figure 1 shown, in the first embodiment of the present invention, a monitoring and control method based on the occurrence law of harmful organisms of pitaya is provided, including:

[0057] S102, retrieve and obtain the historical pest occurrence data of pitaya for data preprocessing. Divide the historical pest occurrence data according to pest categories to generate subsets of pest occurrence data with pest category labels.

[0058] S104, perform feature extraction in the subsets of pest occurrence data, analyze the occurrence cause features corresponding to pest data samples and the causal relationships between the features, and construct a pest attribution path based on the occurrence cause features and the causal relationships.

[0059] S106, obtain multi-source monitoring data in the monitoring area, construct a multi-source monitoring data sequence according to the multi-source monitoring data, calculate the similarity between the multi-source monitoring data sequence and the pest attribution path, and generate a pest early warning for pitaya in the monitoring area through the similarity.

[0060] S108, extract the pest categories and predicted damage conditions corresponding to the pitaya pest early warning, and obtain a combination of control methods according to the pest categories and predicted damage conditions.

[0061] It should be noted that the main pests of pitaya include anthracnose, canker, stem rot, black spot, aphids, thrips, cotton bollworms, Spodoptera litura, Spodoptera exigua, fruit flies, scale insects, ants, snails, slugs, etc. The historical pest occurrence data of pitaya after data cleaning, data dimensionality reduction, normalization and other preprocessing is used to extract pest category keywords in the historical pest occurrence data by keyword extraction, and calculate the semantic similarity between different pest category keywords to reduce the redundancy of category labels caused by differences in pest names. The extracted pest category keywords are refined through the semantic similarity, and the pest category keywords with the same semantic reference are merged to generate de-redundant pest category labels; clustering is performed using the pest category labels to obtain subsets of pest occurrence data under each pest category label, and the number of pest data samples included in different subsets of pest occurrence data is counted, and subsets of pest occurrence data with the number of pest data samples not meeting the preset quantity standard are screened; the SMOTE method is used for upsampling in the screened subsets of pest occurrence data for data balancing processing, and after the sample quantity meets the standard, subsets of pest occurrence data with pest category labels are output.

[0062] Figure 2 The flowchart showing the acquisition of occurrence cause features in the embodiments of the present invention is shown.

[0063] According to the embodiments of the present invention, performing feature extraction in the subsets of pest occurrence data and analyzing the occurrence cause features corresponding to pest data samples specifically includes:

[0064] S202. Obtain the pest occurrence-related data in each pest occurrence data subset to generate a pest data sample. Read the data change range according to the pest data sample, initially select the influencing factors of different pitaya pests through the data change range, and perform variance filtering on the initially selected influencing factors to generate an influencing factor set.

[0065] S204. Perform feature extraction on the influencing factor set through a genetic algorithm. Initialize the influencing factor set to generate an initial population, encode the initial population, train a pest prediction model using the cross-validation method, and calculate the average accuracy as the fitness value of the individual.

[0066] S206. Select a preset number of individuals from the population based on the fitness value according to the elite selection method, and introduce differential calculation to generate the directions of crossover and mutation. Update the population through mutation and differential operations. When the preset minimum fitness threshold is reached, output the influencing factor subset.

[0067] S208. Retrain the pest prediction model on the basis of the influencing factor subset. Use the SVM training model to obtain the mean square error of each influencing factor as the influencing factor score. Further screen the influencing factors according to the influencing factor score to obtain a new influencing factor subset.

[0068] S210. Use the new influencing factor subset to retrain the pest prediction model again until the model obtains the optimal value, then output the final influencing factor subset to obtain the occurrence causal characteristics corresponding to the corresponding pest category label.

[0069] It should be noted that a feature selection method combining variance filtering and improved cross-validation recursive feature elimination based on a genetic algorithm is used to search for and obtain the influencing factor subsets of different pitaya pests. In the genetic algorithm, the fitness value of each individual is calculated through cross-validation. The influencing factor set is divided into n folds, and the pest prediction model is used on n - 1 folds and tested on the remaining fold. After repeated iterations, the average accuracy is calculated. Select a preset number of elite individuals with the highest fitness values from the population according to the elite selection method. Use the idea of difference between individuals to increase the information exchange between elite individuals, clarify the directions of crossover and mutation, for example, use the difference between the sub-optimal solution and the optimal solution and the difference between the third-optimal solution and the optimal solution to obtain the mutated individuals. In the population, tend to the optimal solution among the selected number of features according to the directions of crossover and mutation. Stop the iteration after reaching the minimum fitness threshold. In each generation of the genetic operation, select the individual with the highest fitness value to generate the influencing factor subset.

[0070] It should be noted that the occurrence cause characteristics corresponding to different pest categories are obtained, the information flow paths between the occurrence cause characteristics are initially defined according to the correlation between the occurrence cause characteristics, the two occurrence cause characteristics with qualified correlation are connected, all possible information transmission paths and transmission directions are obtained according to the connection relationships between different characteristics, and the transfer entropy between the occurrence cause characteristic nodes in different information flow paths is calculated in all the information flow paths; the transfer entropy characterizes the information transmission between two system processes through information entropy. The total transfer entropy between the occurrence cause characteristic nodes in the information flow path is obtained, the information flow path with the largest total transfer entropy is selected, and the causal relationship between the occurrence cause characteristics in the pest occurrence process is characterized according to the selected information flow path.

[0071] The critical value intervals of the occurrence cause characteristics corresponding to the occurrence of pitaya pests are obtained according to the historical pest occurrence data, and the pest occurrence probability when the parameters corresponding to the occurrence cause characteristics are within the critical value intervals is judged based on the historical pitaya planting data; the weight information of each occurrence cause characteristic is generated according to the pest occurrence probability, the information flow path corresponding to the causal relationship is obtained, the weight information is matched with the occurrence cause characteristic nodes, and the critical value intervals are marked for the causal relationship pointing to represent the attribute information of the causal relationship pointing, and the pest attribution path is generated by using the marked and matched information flow path; for example, if the overwintering pests in the tree are not cleared in time, when the environmental temperature and humidity rise to the critical value interval, the pests will hatch and cause losses to the pitaya. Moreover, the pathogenic bacteria overwinter on the diseased part or diseased residues in the form of mycelium or conidia, and the conidia of the pathogenic bacteria are spread by wind, rain or insects. When the suitable growth temperature of the pathogenic bacteria reaches the critical value interval (20-30 °C), it is easy to cause anthracnose of pitaya.

[0072] Figure 3 The flowchart of generating the early warning of pitaya pests in the monitoring area in the embodiment of the present invention is shown.

[0073] According to the embodiment of the present invention, the similarity between the multi-source monitoring data sequence and the pest attribution path is calculated, and the early warning of pitaya pests in the monitoring area is generated through the similarity, specifically:

[0074] S302, obtain the occurrence cause characteristics corresponding to different pest categories, aggregate and de-duplicate all the occurrence cause characteristics to generate the monitoring data categories of the pitaya planting area, obtain multi-source monitoring data based on the monitoring data categories, and construct a multi-source monitoring data sequence in combination with time series after preprocessing;

[0075] S304. Based on the pest attribution path, obtain the occurrence cause characteristic nodes falling into the critical value interval according to the multi-source monitoring data sequence, and obtain the highly associated occurrence cause characteristic nodes corresponding to the occurrence cause characteristic nodes falling into the critical value interval according to the connection relationship of the pest attribution path;

[0076] S306. Perform path fitting according to the occurrence cause characteristic nodes falling into the critical value interval and the highly associated occurrence cause characteristic nodes, segment the fitted path, and within each segment, use the Fréchet distance after node weighting as the basis for similarity quantification to calculate the average similarity between the fitted path and the pest attribution path;

[0077] S308. Obtain the pest categories corresponding to the pest attribution paths with an average similarity greater than the preset similarity threshold, obtain the damage status according to the pest occurrence symptoms of the pitaya plants, and generate a pitaya pest warning according to the pest categories and the damage status.

[0078] It should be noted that the fitted path is segmented, and within each segment, obtain the pairs of occurrence cause characteristic nodes (p a1 , q b1 ), (p a2 , q b2 ), …, (p at , q bt ) in the two paths. Let a1 = 1, b1 = 1, at = m, bt = n. p at represents the occurrence cause characteristic node or highly associated occurrence cause characteristic node falling into the critical value interval in the fitted path, and q bt represents the occurrence cause characteristic node in the pest attribution path. m and n respectively represent the path lengths of the fitted path and the pest attribution path. For the length of the pair of occurrence cause characteristic nodes: d represents the Euclidean distance, then the Fréchet distance F within each segment is expressed as F = min ∥D∥. Use the weight information of the occurrence cause characteristics in the pest attribution path to weight the Fréchet distance of the node pairs as the basis for similarity quantification, and obtain the average similarity of each node pair to represent the similarity between the paths.

[0079] It should be noted that the corresponding pest categories and damage status are read through the dragon fruit pest early warning. The monitoring data categories include temperature, rainfall, sunshine, air and soil moisture, acidity and alkalinity, plant images, production files, etc. The change characteristics of the monitoring data are extracted from the multi-source monitoring data sequence. Based on the historical pest occurrence data, the mapping relationship between the multi-source monitoring data and the pest damage status is trained by deep learning methods such as neural networks. The predicted damage status after a preset time is obtained according to the mapping relationship; it is judged whether the predicted damage status exceeds the preset damage status range. When it does not exceed, the green control measures for the pest category are selected through the relevant knowledge graph. The green control measures in the relevant knowledge graph include light trapping, color plate trapping, food trapping, biological control, water and fertilizer improvement, etc. The control examples of each green control measure are retrieved, and the average control time is obtained based on the control examples to generate the ranking of green control measures; the control time is evaluated through the predicted damage status as the control time standard, and a preset number of green control measures that meet the control time standard are selected according to the ranking of the green control measures for combination to generate a control method combination; when it exceeds, the recommended pesticides for the pest category are selected through the relevant knowledge graph, and the corresponding relationship between different application concentrations and the average control time is obtained according to the predicted damage status; the longest control time is obtained according to the control time standard, and the application concentration of the longest control time is obtained through the corresponding relationship. The recommended pesticides and application concentrations are generated into chemical control measures and combined with the green control measure with the shortest average control time to generate a control method combination.

[0080] The production files during the dragon fruit planting process are obtained through the multi-source monitoring data of the monitoring area. The generated files include measures such as water and fertilizer management, pest control, pruning, and harvesting. The influence degree of the current planting method on the dragon fruit pests is evaluated according to the production files, and the pest occurrence probability of the dragon fruit is predicted according to the production files. The pest occurrence probability is sent to the grower terminal in a preset manner. The current cultivation management measures required in the monitoring area are obtained through the multi-source monitoring data, and the grower is urged to implement the cultivation management measures through data monitoring. If the grower fails to implement the cultivation management measures within the preset time period, the monitoring frequency is increased, and the pest occurrence probability of the dragon fruit is updated in a timely manner for pushing, and the extensive management planting method of the grower is gradually changed by using the improved cultivation management means.

[0081] Figure 4 The block diagram of the monitoring and control system based on the occurrence law of dragon fruit pests of the present invention is shown.

[0082] In the second aspect of the present invention, a monitoring and control system 4 based on the occurrence law of harmful organisms in pitaya is provided, including a harmful organism occurrence law analysis module 41, a multi-source monitoring data acquisition module 42, a pitaya harmful organism early warning module 43, a control decision-making module 44, and a data visualization module 45;

[0083] The harmful organism occurrence law analysis module is responsible for analyzing the occurrence cause characteristics corresponding to different harmful organism categories and the causal relationships between the characteristics based on historical harmful organism occurrence data, constructing corresponding harmful organism attribution paths, and characterizing the occurrence law of harmful organisms in pitaya;

[0084] The multi-source monitoring data acquisition module is responsible for collecting multi-source monitoring data in the monitoring area, performing serialization processing, and constructing a multi-source monitoring data sequence with the same structure as the harmful organism attribution path;

[0085] The pitaya harmful organism early warning module is responsible for calculating the similarity between the multi-source monitoring data sequence and different harmful organism attribution paths, and generating a pitaya harmful organism early warning based on the similarity judgment;

[0086] The control decision-making module is responsible for obtaining a combination of control methods dominated by green control measures based on the early warning harmful organism category and predicted harm status;

[0087] The data visualization module is responsible for displaying the combination of control methods in the monitoring area, and performing real-time visualization display on the multi-source monitoring data and the pitaya harmful organism early warning.

[0088] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the displayed or discussed components can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0089] The units described as separate components above may or may not be physically separated, and the components displayed as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0090] In addition, each functional unit in the embodiments of the present invention can be all integrated into one processing unit, or each unit can be separately used as one unit alone, or two or more units can be integrated into one unit; the above integrated unit can be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0091] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium includes: removable storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks and other various media that can store program codes.

[0092] Alternatively, if the above integrated unit of the present invention is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. And the foregoing storage medium includes: removable storage devices, ROM, RAM, magnetic disks, or optical disks and other various media that can store program codes.

[0093] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, and all of them should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A monitoring and control method based on the occurrence law of dragon fruit harmful organisms, characterized in that, The following steps are involved: Retrieving and obtaining historical pest occurrence data of pitaya for data preprocessing, dividing the historical pest occurrence data according to pest categories, and generating a subset of pest occurrence data with pest category labels; Extracting features from the pest occurrence data subset, analyzing occurrence causal features corresponding to the pest data sample and the causal relationship between the features, and constructing a pest attribution path based on the occurrence causal features and the causal relationship; Acquire multi-source monitoring data in the monitoring area, construct a multi-source monitoring data sequence according to the multi-source monitoring data, use Fréchet distance to calculate the similarity between the multi-source monitoring data sequence and the pest attribution path, and generate a pitaya pest warning in the monitoring area according to the similarity; Extract the pest categories and predicted damage conditions corresponding to the dragon fruit pest warning, and obtain the combination of prevention and control methods based on the pest categories and predicted damage conditions; The pest attribution path is constructed based on the occurrence causal characteristics and the causal relationship, specifically: Obtain the critical value interval of the corresponding causal characteristics when pitaya pests occur according to the historical pest occurrence data, and determine the probability of pest occurrence when the corresponding parameters of the causal characteristics are within the critical value interval based on the historical pitaya planting data; Generate weight information of each occurrence causal feature according to the occurrence probability of the pest, obtain the information flow path corresponding to the causal relationship, match the weight information with the occurrence causal feature node, and mark the critical value interval of the causal relationship to characterize the attribute information of the causal relationship; The pest attribution path is generated using the information flow path after tag matching.

2. A monitoring and control method based on the occurrence law of dragon fruit harmful organisms according to claim 1, characterized in that, The historical pest occurrence data is divided according to pest categories to generate a subset of pest occurrence data with pest category labels, specifically: Acquire historical pest occurrence data after data cleaning and normalization, use keyword extraction to obtain pest category keywords in the historical pest occurrence data, and calculate the semantic similarity between different pest category keywords; The extracted pest category keywords are simplified by using the semantic similarity, and pest category keywords with the same semantics are merged to generate pest category labels after redundancy removal; Clustering is performed using the pest category labels to obtain pest occurrence data subsets under each pest category label, and the number of pest data samples contained in different pest occurrence data subsets is counted to screen pest occurrence data subsets whose number of pest data samples does not meet a preset number standard; Upsampling is performed on the pest occurrence data subset obtained through screening, and data balancing is performed. After the sample quantity reaches the standard, the pest occurrence data subset with pest category labels is output.

3. A monitoring and control method based on the occurrence law of dragon fruit harmful organisms according to claim 1, characterized in that, Feature extraction is performed in the pest occurrence data subset, and the occurrence cause characteristics corresponding to the pest data samples are analyzed, specifically: Acquire pest occurrence related data in each pest occurrence data subset to generate pest data samples, read the data variation range according to the pest data samples, preliminarily select the influencing factors of different dragon fruit pests according to the data variation range, and perform variance filtering on the preliminarily selected influencing factors to generate an influencing factor set; Extracting features from the influencing factor set by a genetic algorithm, initializing the influencing factor set to generate an initial population, encoding the initial population, training the pest prediction model by a cross-validation method, and calculating the average accuracy as the fitness value of the individual; Based on the fitness value, a preset number of individuals are selected from the population according to an elite selection method, and differential calculation is introduced to generate the direction of crossover and mutation, the population is updated through mutation and differential operations, and a subset of influencing factors is output when a preset minimum fitness threshold is reached; Retraining the pest prediction model based on the influencing factor subset, using the SVM training model to obtain the mean square error of each influencing factor as an influencing factor score, further screening the influencing factors according to the influencing factor score, and obtaining a new influencing factor subset; The pest prediction model is trained again using the new subset of influencing factors until the model obtains the optimal value, and then the final subset of influencing factors is output to obtain the occurrence causal characteristics corresponding to the corresponding pest category label.

4. A monitoring and control method based on the occurrence law of dragon fruit harmful organisms according to claim 3, characterized in that, Obtain the causal relationship between the occurrence causal characteristics, specifically: Obtain the occurrence causal characteristics corresponding to different pest categories, initially define the information flow path between the occurrence causal characteristics according to the correlation between the occurrence causal characteristics, and calculate the transfer entropy between each occurrence causal characteristic node in different information flow paths; The sum of the transfer entropies between the causal feature nodes in the information flow path is obtained, and the information flow path with the largest sum of transfer entropies is selected. The causal relationship between the causal features in the process of pest occurrence is characterized according to the selected information flow path.

5. A monitoring and control method based on the occurrence law of dragon fruit harmful organisms according to claim 1, characterized in that, The similarity between the multi-source monitoring data sequence and the pest attribution path is calculated, and the pitaya pest warning in the monitoring area is generated through the similarity, specifically: Obtain the occurrence causal characteristics corresponding to different pest categories, aggregate and remove redundancy of all occurrence causal characteristics, generate monitoring data categories of pitaya planting areas, obtain multi-source monitoring data based on the monitoring data categories, and construct a multi-source monitoring data sequence after pre-processing and combining with time series; Based on the pest attribution path, the occurrence causative feature nodes falling into the critical value interval are obtained according to the multi-source monitoring data sequence, and the highly correlated occurrence causative feature nodes corresponding to the occurrence causative feature nodes falling into the critical value interval are obtained according to the connection relationship of the pest attribution path; Path fitting is performed based on the occurrence causal characteristic nodes and the highly correlated occurrence causal characteristic nodes that fall into the critical value interval, and the fitted path is segmented. In each segment, the weighted Fréchet distance of the node is used as the basis for similarity quantification, and the average similarity between the fitted path and the pest attribution path is calculated; The pest category corresponding to the pest attribution path with an average similarity greater than a preset similarity threshold is obtained, and the damage status is obtained according to the symptoms of the pests in the pitaya plant, and a pitaya pest warning is generated according to the pest category and the damage status.

6. A monitoring and control method based on the occurrence law of dragon fruit harmful organisms according to claim 1, characterized in that, Obtain a combination of control methods based on pest categories and predicted damage conditions, specifically: The corresponding pest category and hazard status are read through the dragon fruit pest early warning, the monitoring data change characteristics are extracted according to the multi-source monitoring data sequence, the mapping relationship between the multi-source monitoring data and the pest hazard status is trained using the historical pest occurrence data, and the predicted hazard status after a preset time is obtained according to the mapping relationship; According to the predicted hazard status, it is judged whether it exceeds the preset hazard status range. If it does not exceed, the green control measures of the pest category are selected through the relevant knowledge graph, and the prevention and control examples of each green control measure are retrieved and obtained. Based on the prevention and control examples, the average prevention and control time is obtained, and the ranking of green control measures is generated; By predicting the hazard status and evaluating the prevention and control time as the prevention and control time standard, a preset number of green prevention and control measures that meet the prevention and control time standard are selected according to the green prevention and control measures ranking to combine and generate a prevention and control method combination; When it exceeds the limit, the recommended pesticides for the pest category are selected through the relevant knowledge graph, and the corresponding relationship between different pesticide concentrations and average control time is obtained based on the predicted hazard status; The longest control time is obtained according to the control time standard, and the drug concentration for the longest control time is obtained through the corresponding relationship. The recommended agents and drug concentrations are used to generate chemical control measures, which are combined with green control measures with the shortest average control time to generate a control method combination.

7. A monitoring and control system based on the occurrence law of dragon fruit pests, realizing the monitoring and control method based on the occurrence law of dragon fruit pests as described in any one of claims 1-6, characterized in that, It includes the pest occurrence law analysis module, multi-source monitoring data collection module, dragon fruit pest early warning module, prevention and control decision module and data visualization module; The pest occurrence law analysis module is responsible for analyzing the occurrence causal characteristics corresponding to different pest categories and the causal relationship between the characteristics according to the historical pest occurrence data, constructing the corresponding pest attribution path, and characterizing the occurrence law of pitaya pests; The multi-source monitoring data acquisition module is responsible for collecting multi-source monitoring data of the monitoring area, and performing serialization processing to construct a multi-source monitoring data sequence with the same structure as the pest attribution path; The pitaya pest early warning module is responsible for calculating the similarity between the multi-source monitoring data sequence and the attribution paths of different pests, and generating a pitaya pest early warning based on the similarity judgment; The control decision module is responsible for obtaining a combination of control methods dominated by green control measures based on the early warning pest categories and predicted hazard conditions; The data visualization module is responsible for displaying the combination of prevention and control methods in the monitored area, and for real-time visualization of multi-source monitoring data and dragon fruit pest warning.

Images