Forestry disease and insect pest intelligent decision-making method and system based on multi-modal knowledge graph

Abstract

The embodiment of the invention relates to the technical field of forestry disease and pest control, and particularly discloses a forestry disease and pest intelligent decision-making method and system based on a multi-modal knowledge graph. According to the embodiment of the invention, the method comprises the steps: obtaining the multi-modal forestry data of a target forestry region, and carrying out the preprocessing of the multi-modal forestry data; entity detection, relation extraction and entity alignment processing are carried out, and a multi-modal information vector is constructed; constructing a multi-modal knowledge graph; performing association identification and propagation reasoning on forestry diseases and pests to obtain a disease and pest reasoning result; and according to the disease and pest reasoning result, forestry disease and pest decision processing is performed. Multi-modal forestry data can be acquired, multi-modal information vectors can be constructed, then a multi-modal knowledge graph can be constructed, association identification and propagation reasoning of forestry diseases and insect pests can be carried out, and dynamic treatment decision processing can be carried out, so that the multi-modal knowledge graph is applied to an intelligent decision-making process of the forestry diseases and insect pests, the disease and insect pest identification is more accurate, and the decision-making efficiency is improved. And the treatment efficiency of forestry diseases and insect pests is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of forestry pest control, and in particular relates to a forestry pest intelligent decision-making method and system based on a multimodal knowledge graph. Background Art

[0002] Forestry pests and diseases not only destroy the growth of trees, reduce forest quality and economic value, but may also cause ecological imbalance, forest degradation and even biodiversity loss.

[0003] Forestry pest and disease control is a comprehensive prevention and control measure taken against various harmful organisms occurring in forest ecosystems to protect the health and sustainable development of forest resources.

[0004] Forestry pest and disease control is not only a key link in ensuring forest health, but also a basic work to maintain ecological security, promote carbon sequestration capacity and promote green development.

[0005] Existing technologies for forest pest and disease control only monitor and control through a simple and single method. It is difficult to integrate multimodal information such as remote sensing, meteorology, text, and monitoring, resulting in incomplete or distorted identification of pests and diseases. Traditional control methods are often based on historical experience or static rules and cannot make dynamic decisions based on specific environments, transmission trends, and risk levels, resulting in low control efficiency. Summary of the Invention

[0006] The purpose of the embodiments of the present invention is to provide a forestry pest and disease intelligent decision-making method and system based on a multimodal knowledge graph, aiming to solve the problems raised in the background technology.

[0007] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0008] An intelligent decision-making method for forestry pests and diseases based on a multimodal knowledge graph, the method specifically comprises the following steps:

[0009] Determining a target forestry area, obtaining multimodal forestry data of the target forestry area, and preprocessing the multimodal forestry data to obtain multimodal cleaned data;

[0010] Performing entity detection, relationship extraction, and entity alignment on the multimodal cleansed data to construct a multimodal information vector;

[0011] Constructing a multimodal knowledge graph based on the multimodal information vector;

[0012] Based on the multimodal knowledge graph, association identification and propagation reasoning of forest pests and diseases are performed to obtain pest and disease reasoning results;

[0013] Decision-making and treatment of forestry pests and diseases are carried out based on the pest and disease reasoning results.

[0014] As a further limitation of the technical solution of the embodiment of the present invention, determining the target forestry area, obtaining multimodal forestry data of the target forestry area, and preprocessing the multimodal forestry data to obtain multimodal cleaned data specifically include the following steps:

[0015] Identify target forestry areas and multiple data types;

[0016] Performing multimodal data collection on the target forestry area according to the plurality of data types to obtain multimodal forestry data;

[0017] performing image enhancement and segmentation on the multimodal forestry data;

[0018] Performing text cleaning on the multimodal forestry data;

[0019] performing environmental data normalization on the multimodal forestry data;

[0020] Generate multimodal cleaned data.

[0021] As a further limitation of the technical solution of the embodiment of the present invention, the performing entity detection, relationship extraction, and entity alignment processing on the multimodal cleansed data to construct a multimodal information vector specifically includes the following steps:

[0022] Performing entity detection on the multimodal cleansed data to determine multiple target entities;

[0023] Based on the plurality of target entities, performing relationship extraction on the multimodal cleansed data and recording entity relationship information;

[0024] Performing multimodal entity alignment on the multimodal cleaned data and recording the entity alignment result;

[0025] A multimodal information vector is constructed according to the entity relationship information, the entity alignment result and the multiple target entities.

[0026] As a further limitation of the technical solution of the embodiment of the present invention, constructing a multimodal knowledge graph based on the multimodal information vector specifically includes the following steps:

[0027] constructing a plurality of structured triples according to the multimodal information vector;

[0028] Use graph database for management;

[0029] Performing graph embedding on a plurality of the structured triples;

[0030] Build a multimodal knowledge graph that integrates images, text, and environment.

[0031] As a further limitation of the technical solution of the embodiment of the present invention, the association identification and propagation reasoning of forestry pests and diseases based on the multimodal knowledge graph and obtaining the pest and disease reasoning results specifically include the following steps:

[0032] Aggregating graph structure information in the multimodal knowledge graph;

[0033] Performing association identification of forestry pests and diseases on the atlas structure information;

[0034] Combined with the preset Bayesian path, the source and transmission path of the disease are predicted;

[0035] Obtain pest and disease inference results.

[0036] As a further limitation of the technical solution of the embodiment of the present invention, the decision-making process for forestry pests and diseases based on the pest and disease reasoning result specifically includes the following steps:

[0037] Based on the pest and disease reasoning results, decision-making planning for forest pests and diseases is carried out to obtain multiple pest and disease control strategies;

[0038] Perform combined optimization calculations on multiple pest control strategies and select multiple target control strategies;

[0039] Divide into multiple pest and disease control areas;

[0040] Calculating management priority scores for the plurality of pest and disease management areas;

[0041] Matching multiple corresponding governance priorities according to the multiple governance priority scores;

[0042] According to the plurality of control priorities and the plurality of target control strategies, batch control of forestry pests and diseases is carried out in the plurality of pest and disease control areas.

[0043] As a further limitation of the technical solution of the embodiment of the present invention, the objective function of the combinatorial optimization calculation is:

[0044]

[0045] Among them, R is the total revenue, x i represents the selection state of the i-th pest control strategy, x i 1 represents selection, x i 0 means no selection, there are n pest control strategies, b i is the management benefit of the i-th pest control strategy, a i is the management cost of the i-th pest control strategy;

[0046] The calculation formula for multiple governance priorities is:

[0047] P j =k1H j +k2V j +k3D j ;

[0048] k1+k2+k3=1;

[0049] Among them, P j represents the management priority of the jth pest control area, H j represents the risk score of the jth pest control area, V j represents the forestry value of the jth pest control area, D j represents the pest density in the jth pest control area, and k1, k2, and k3 are preset weight factors.

[0050] An intelligent decision-making system for forestry pests and diseases based on a multimodal knowledge graph, comprising a multimodal data acquisition unit, an entity detection and analysis unit, a knowledge graph construction unit, a forestry pest and disease reasoning unit, and a pest and disease decision-making processing unit, wherein:

[0051] a multimodal data acquisition unit, configured to determine a target forestry area, acquire multimodal forestry data of the target forestry area, and preprocess the multimodal forestry data to acquire multimodal cleaned data;

[0052] An entity detection and analysis unit, configured to perform entity detection, relationship extraction, and entity alignment processing on the multimodal cleansed data to construct a multimodal information vector;

[0053] A knowledge graph construction unit, configured to construct a multimodal knowledge graph based on the multimodal information vector;

[0054] A forestry pest and disease reasoning unit, configured to perform association identification and propagation reasoning of forestry pests and diseases based on the multimodal knowledge graph, and obtain pest and disease reasoning results;

[0055] The pest and disease decision processing unit is used to make decisions on forest pests and diseases based on the pest and disease reasoning results.

[0056] As a further limitation of the technical solution of the embodiment of the present invention, the multimodal data acquisition unit specifically includes:

[0057] an area determination module for determining target forestry areas and multiple data types;

[0058] A multimodal data acquisition module, configured to perform multimodal data acquisition on the target forestry area according to the plurality of data types to obtain multimodal forestry data;

[0059] An image enhancement module, configured to perform image enhancement and segmentation on the multimodal forestry data;

[0060] A text cleaning module, used for performing text cleaning on the multimodal forestry data;

[0061] A normalization processing module, used for performing environmental data normalization on the multimodal forestry data;

[0062] The multimodal clean data generation module is used to generate multimodal clean data.

[0063] As a further limitation of the technical solution of the embodiment of the present invention, the pest and disease decision processing unit specifically includes:

[0064] A decision-making and planning module is used to make decision-making plans for forestry pests and diseases based on the pest and disease reasoning results and obtain multiple pest and disease control strategies;

[0065] Combinatorial optimization calculation module, used to perform combinatorial optimization calculations on multiple pest control strategies and select multiple target control strategies;

[0066] Regional division module, used to divide multiple pest and disease control areas;

[0067] A control priority score calculation module, used to calculate the control priority scores of the plurality of pest control areas;

[0068] a priority matching module, configured to match a plurality of corresponding governance priorities according to the plurality of governance priority scores;

[0069] The batch control module is used to carry out batch control of forest pests and diseases in multiple pest control areas according to multiple control priorities and multiple target control strategies.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] The embodiment of the present invention obtains multimodal forestry data of a target forestry area and preprocesses the multimodal forestry data; performs entity detection, relationship extraction, and entity alignment processing to construct a multimodal information vector; constructs a multimodal knowledge graph; performs association identification and propagation reasoning of forestry pests and diseases to obtain pest and disease reasoning results; and performs forestry pest and disease decision-making based on the pest and disease reasoning results. The system can obtain multimodal forestry data, construct a multimodal information vector, and then construct a multimodal knowledge graph to perform association identification and propagation reasoning of forestry pests and diseases, and perform dynamic management decision-making processing, thereby applying the multimodal knowledge graph to the intelligent decision-making process of forestry pests and diseases, making pest and disease identification more accurate and effectively improving the management efficiency of forestry pests and diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention.

[0073] Figure 1 A flow chart of a method provided by an embodiment of the present invention is shown.

[0074] Figure 2 A flow chart of obtaining multimodal cleaning data in the method provided by an embodiment of the present invention is shown.

[0075] Figure 3 A flowchart of constructing a multimodal information vector in the method provided by an embodiment of the present invention is shown.

[0076] Figure 4 A flowchart of constructing a multimodal knowledge graph in the method provided in an embodiment of the present invention is shown.

[0077] Figure 5 A flow chart of obtaining pest and disease reasoning results in the method provided by an embodiment of the present invention is shown.

[0078] Figure 6 The flowchart of the decision-making process for forest pests and diseases in the method provided by the embodiment of the present invention is shown.

[0079] Figure 7 The application architecture diagram of the system provided by the embodiment of the present invention is shown.

[0080] Figure 8 The figure shows a structural block diagram of a multimodal data acquisition unit in a system provided by an embodiment of the present invention.

[0081] Figure 9 The figure shows a structural block diagram of a pest and disease decision processing unit in a system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0082] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0083] It is understandable that the existing technology for forest pest control only monitors and controls in a simple and single way, and it is difficult to integrate multimodal information such as remote sensing, meteorology, text, and monitoring, resulting in incomplete or distorted identification of pests and diseases. In addition, traditional control methods are often based on historical experience or static rules, and cannot make dynamic decisions based on specific environments, transmission trends, and risk levels, resulting in low control efficiency.

[0084] To solve the above problems, the embodiment of the present invention determines a target forestry area, obtains multimodal forestry data of the target forestry area, and pre-processes the multimodal forestry data to obtain multimodal cleaned data; performs entity detection, relationship extraction and entity alignment on the multimodal cleaned data to construct a multimodal information vector; constructs a multimodal knowledge graph based on the multimodal information vector; performs association identification and propagation reasoning of forestry pests and diseases based on the multimodal knowledge graph to obtain pest and disease reasoning results; and performs decision-making on forestry pests and diseases based on the pest and disease reasoning results. It is possible to obtain multimodal forestry data, construct a multimodal information vector, and then construct a multimodal knowledge graph to perform association identification and propagation reasoning of forestry pests and diseases, and perform dynamic management decision-making processing, thereby applying the multimodal knowledge graph to the intelligent decision-making process of forestry pests and diseases, making pest and disease identification more accurate and effectively improving the management efficiency of forestry pests and diseases.

[0085] Figure 1 A flow chart of a method provided by an embodiment of the present invention is shown.

[0086] Specifically, the forestry pest and disease intelligent decision-making method based on multimodal knowledge graph includes the following steps:

[0087] Step S101: determine a target forestry area, obtain multimodal forestry data of the target forestry area, and preprocess the multimodal forestry data to obtain multimodal cleaned data.

[0088] In an embodiment of the present invention, by determining a target forestry area and multiple data types, and then performing multimodal data collection on the target forestry area according to the multiple data types, multimodal forestry data (including image data, time series sensor data, text data, and spatial data, etc.) is obtained, and then the multimodal forestry data is preprocessed by image enhancement and segmentation, text cleaning (stop word removal, word segmentation, spelling correction, syntactic analysis, etc.) and environmental data normalization to generate multimodal cleaned data.

[0089] It can be understood that image data includes leaf images, pest images, etc.; time series sensor data includes temperature, humidity, wind speed, etc.; text data includes forestry reports and disease diagnosis descriptions, etc.; spatial data includes remote sensing images and GPS positioning, etc.

[0090] Specifically, Figure 2 A flow chart of obtaining multimodal cleaning data in the method provided by an embodiment of the present invention is shown.

[0091] Among them, in the preferred embodiment provided by the present invention, the determining of the target forestry area, obtaining multimodal forestry data of the target forestry area, and preprocessing the multimodal forestry data to obtain multimodal cleaned data specifically include the following steps:

[0092] Step S1011, determining a target forestry area and multiple data types;

[0093] Step S1012: performing multimodal data collection on the target forestry area according to the plurality of data types to obtain multimodal forestry data;

[0094] Step S1013, performing image enhancement and segmentation on the multimodal forestry data;

[0095] Step S1014, performing text cleaning on the multimodal forestry data;

[0096] Step S1015, performing environmental data normalization on the multimodal forestry data;

[0097] Step S1016: Generate multimodal cleaning data.

[0098] Furthermore, the forestry pest and disease intelligent decision-making method based on the multimodal knowledge graph further includes the following steps:

[0099] Step S102 : performing entity detection, relationship extraction, and entity alignment processing on the multimodal cleansed data to construct a multimodal information vector.

[0100] In an embodiment of the present invention, entity detection is performed on the multimodal clean data (using algorithms such as YOLO and Mask-RCNN for target detection and segmentation), multiple target entities are determined, and then based on the multiple target entities, relationship extraction is performed on the multimodal clean data (extracting relationships such as "pests-impacts-tree species"), the entity relationship information is recorded, and multimodal entity alignment is performed on the multimodal clean data (matching entities in images and texts), and the entity alignment results are recorded. Then, a multimodal information vector is constructed according to the entity relationship information, the entity alignment results, and the multiple target entities.

[0101] Specifically, Figure 3 A flowchart of constructing a multimodal information vector in the method provided by an embodiment of the present invention is shown.

[0102] In a preferred embodiment of the present invention, performing entity detection, relationship extraction, and entity alignment on the multimodal cleansed data to construct a multimodal information vector specifically includes the following steps:

[0103] Step S1021: performing entity detection on the multimodal cleansed data to determine multiple target entities;

[0104] Step S1022: extracting relationships from the multimodal cleansed data based on the plurality of target entities, and recording entity relationship information;

[0105] Step S1023: performing multimodal entity alignment on the multimodal cleaned data and recording the entity alignment result;

[0106] Step S1024: construct a multimodal information vector according to the entity relationship information, the entity alignment result and the multiple target entities.

[0107] Furthermore, the forestry pest and disease intelligent decision-making method based on the multimodal knowledge graph further includes the following steps:

[0108] Step S103: construct a multimodal knowledge graph based on the multimodal information vector.

[0109] In an embodiment of the present invention, entities and relationships are constructed into multiple structured triples based on multimodal information vectors, and a graph database is used for management. The multiple structured triples are embedded in a graph to construct a multimodal knowledge graph that integrates images, texts, and environments.

[0110] Specifically, Figure 4 A flowchart of constructing a multimodal knowledge graph in the method provided in an embodiment of the present invention is shown.

[0111] In a preferred embodiment of the present invention, the step of constructing a multimodal knowledge graph based on the multimodal information vector specifically includes the following steps:

[0112] Step S1031, constructing a plurality of structured triples according to the multimodal information vector;

[0113] Step S1032: Use a graph database for management;

[0114] Step S1033, performing graph embedding on the plurality of structured triples;

[0115] Step S1034: construct a multimodal knowledge graph that integrates images, text, and environment.

[0116] Furthermore, the forestry pest and disease intelligent decision-making method based on the multimodal knowledge graph further includes the following steps:

[0117] Step S104: Based on the multimodal knowledge graph, association identification and propagation reasoning of forest pests and diseases are performed to obtain pest and disease reasoning results.

[0118] In an embodiment of the present invention, the graph structure information in the multimodal knowledge graph is aggregated, and the association of forestry pests and diseases is identified on the graph structure information. Combined with the preset Bayesian path, the source and transmission path of the disease are predicted to obtain the pest and disease reasoning results.

[0119] Specifically, Figure 5A flow chart of obtaining pest and disease reasoning results in the method provided by an embodiment of the present invention is shown.

[0120] Among them, in the preferred embodiment provided by the present invention, the association identification and propagation reasoning of forestry pests and diseases based on the multimodal knowledge graph and the acquisition of the pest and disease reasoning results specifically include the following steps:

[0121] Step S1041, aggregating graph structure information in the multimodal knowledge graph;

[0122] Step S1042, performing association identification of forestry pests and diseases on the graph structure information;

[0123] Step S1043: predicting the source and transmission path of the disease in combination with the preset Bayesian path;

[0124] Step S1044: Obtain the pest and disease inference result.

[0125] Furthermore, the forestry pest and disease intelligent decision-making method based on the multimodal knowledge graph further includes the following steps:

[0126] Step S105: Decision-making on forestry pests and diseases is performed based on the pest and disease reasoning result.

[0127] In an embodiment of the present invention, decision-making planning for forestry pests and diseases is performed based on the pest and disease reasoning results, multiple pest and disease control strategies are obtained, and a combination optimization calculation is performed on the multiple pest and disease control strategies according to the objective function of the combination optimization calculation. Multiple target control strategies are selected from the multiple pest and disease control strategies, and the target forestry area is divided into multiple pest and disease control areas. The control priority scores of the multiple pest and disease control areas are calculated, and then, based on the multiple control priority scores, multiple corresponding control priorities are matched. Then, according to the multiple control priorities and the multiple target control strategies, batch control of forestry pests and diseases is performed on the multiple pest and disease control areas. Specifically, the objective function of the combination optimization calculation is:

[0128]

[0129] Among them, R is the total revenue, x i represents the selection state of the i-th pest control strategy, x i 1 represents selection, x i 0 means no selection, there are n pest control strategies, b i is the management benefit of the i-th pest control strategy, a i is the management cost of the i-th pest control strategy;

[0130] The calculation formula for multiple governance priorities is:

[0131] Pj =k1H j +k2V j +k3D j ;

[0132] k1+k2+k3=1;

[0133] Among them, P j represents the management priority of the jth pest control area, H j represents the risk score of the jth pest control area, V j represents the forestry value of the jth pest control area, D j represents the pest density in the jth pest control area, and k1, k2, and k3 are preset weight factors.

[0134] Specifically, Figure 6 The flowchart of the decision-making process for forest pests and diseases in the method provided by the embodiment of the present invention is shown.

[0135] Among them, in the preferred embodiment provided by the present invention, the decision-making process for forestry pests and diseases based on the pest and disease reasoning result specifically includes the following steps:

[0136] Step S1051: Based on the pest inference results, decision-making planning for forest pests and diseases is performed to obtain multiple pest and disease control strategies;

[0137] Step S1052: performing combined optimization calculations on multiple pest control strategies and selecting multiple target control strategies;

[0138] Step S1053, dividing into multiple pest control areas;

[0139] Step S1054, calculating the control priority scores of the plurality of pest control areas;

[0140] Step S1055: Matching multiple corresponding governance priorities according to the multiple governance priority scores;

[0141] Step S1056: Batch control of forestry pests and diseases is performed on the plurality of pest and disease control areas according to the plurality of control priorities and the plurality of target control strategies.

[0142] Further, Figure 7 The application architecture diagram of the system provided by the embodiment of the present invention is shown.

[0143] Among them, in another preferred embodiment provided by the present invention, the forestry pest and disease intelligent decision-making system based on the multimodal knowledge graph includes:

[0144] The multimodal data acquisition unit 101 is used to determine a target forestry area, acquire multimodal forestry data of the target forestry area, and preprocess the multimodal forestry data to acquire multimodal cleaned data.

[0145] In an embodiment of the present invention, the multimodal data acquisition unit 101 determines a target forestry area and multiple data types, and then collects multimodal data for the target forestry area according to the multiple data types to obtain multimodal forestry data (including image data, time series sensor data, text data, and spatial data, etc.), and then performs image enhancement and segmentation, text cleaning (stop word removal, word segmentation, spelling correction, syntactic analysis, etc.) and environmental data normalization preprocessing on the multimodal forestry data to generate multimodal cleaned data.

[0146] Specifically, Figure 8 FIG. 1 shows a structural block diagram of the multimodal data acquisition unit 101 in the system provided by an embodiment of the present invention.

[0147] In a preferred embodiment of the present invention, the multimodal data acquisition unit 101 specifically includes:

[0148] The region determination module 1011 is used to determine a target forestry region and multiple data types;

[0149] A multimodal data collection module 1012 is configured to collect multimodal data of the target forestry area according to the plurality of data types to obtain multimodal forestry data;

[0150] An image enhancement module 1013 is used to perform image enhancement and segmentation on the multimodal forestry data;

[0151] A text cleaning module 1014 is used to perform text cleaning on the multimodal forestry data;

[0152] A normalization processing module 1015 is used to perform environmental data normalization on the multimodal forestry data;

[0153] The multimodal cleaned data generating module 1016 is configured to generate multimodal cleaned data.

[0154] Furthermore, the forestry pest and disease intelligent decision-making system based on the multimodal knowledge graph also includes:

[0155] The entity detection and analysis unit 102 is used to perform entity detection, relationship extraction and entity alignment processing on the multimodal cleansed data to construct a multimodal information vector.

[0156] In an embodiment of the present invention, the entity detection and analysis unit 102 determines multiple target entities by performing entity detection on the multimodal clean data (using algorithms such as YOLO and Mask-RCNN for target detection and segmentation), and then extracts relationships on the multimodal clean data based on the multiple target entities (extracting relationships such as "pests-impacts-tree species"), records the entity relationship information, and performs multimodal entity alignment on the multimodal clean data (matching entities in images and texts), records the entity alignment results, and then constructs a multimodal information vector according to the entity relationship information, entity alignment results, and multiple target entities.

[0157] The knowledge graph construction unit 103 is used to construct a multimodal knowledge graph based on the multimodal information vector.

[0158] In an embodiment of the present invention, the knowledge graph construction unit 103 constructs multiple structured triples of entities and relationships based on the multimodal information vector, manages them using a graph database, and embeds the multiple structured triples into a graph to construct a multimodal knowledge graph that integrates images, texts, and environments.

[0159] The forestry pest and disease reasoning unit 104 is used to perform association identification and propagation reasoning of forestry pests and diseases based on the multimodal knowledge graph to obtain pest and disease reasoning results.

[0160] In an embodiment of the present invention, the forestry pest and disease reasoning unit 104 aggregates the graph structure information in the multimodal knowledge graph, and performs association identification of forestry pests and diseases on the graph structure information, and combines the preset Bayesian path to predict the source and transmission path of the disease to obtain the pest and disease reasoning result.

[0161] The pest and disease decision processing unit 105 is used to make a decision on forest pests and diseases based on the pest and disease reasoning result.

[0162] In an embodiment of the present invention, the pest decision processing unit 105 performs decision-making planning for forest pests based on the pest reasoning results, obtains multiple pest control strategies, and performs combined optimization calculation on the multiple pest control strategies according to the objective function of the combined optimization calculation. From the multiple pest control strategies, multiple target control strategies are selected, and the target forestry area is divided into multiple pest control areas. The control priority scores of the multiple pest control areas are calculated, and then, based on the multiple control priority scores, multiple corresponding control priorities are matched. Then, according to the multiple control priorities and the multiple target control strategies, batch control of forest pests is performed on the multiple pest control areas. Specifically, the objective function of the combined optimization calculation is:

[0163]

[0164] Among them, R is the total revenue, x i represents the selection state of the i-th pest control strategy, x i 1 represents selection, x i 0 means no selection, there are n pest control strategies, b i is the management benefit of the i-th pest control strategy, a i is the management cost of the i-th pest control strategy;

[0165] The calculation formula for multiple governance priorities is:

[0166] P j =k1H j +k2V j +k3D j ;

[0167] k1+k2+k3=1;

[0168] Among them, P j represents the management priority of the jth pest control area, H j represents the risk score of the jth pest control area, V j represents the forestry value of the jth pest control area, D j represents the pest density in the jth pest control area, and k1, k2, and k3 are preset weight factors.

[0169] Specifically, Figure 9 FIG. 1 shows a structural block diagram of the pest and disease decision processing unit 105 in the system provided by an embodiment of the present invention.

[0170] Among them, in the preferred embodiment provided by the present invention, the pest and disease decision processing unit 105 specifically includes:

[0171] A decision-making and planning module 1051 is used to perform decision-making and planning on forestry pests and diseases based on the pest and disease reasoning results, and obtain multiple pest and disease control strategies;

[0172] Combination optimization calculation module 1052, used to perform combination optimization calculation on multiple pest control strategies and select multiple target control strategies;

[0173] The area division module 1053 is used to divide multiple pest control areas;

[0174] A control priority score calculation module 1054 is used to calculate control priority scores for the plurality of pest control areas;

[0175] A priority matching module 1055 is configured to match a plurality of corresponding governance priorities according to the plurality of governance priority scores;

[0176] The batch control module 1056 is used to perform batch control of forest pests and diseases in the plurality of pest and disease control areas according to the plurality of control priorities and the plurality of target control strategies.

[0177] It should be understood that, although the various steps in the flow chart of each embodiment of the present invention are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0178] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0179] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0180] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0181] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph, characterized by: The method specifically comprises the following steps: Determining a target forestry area, obtaining multimodal forestry data of the target forestry area, and preprocessing the multimodal forestry data to obtain multimodal cleaned data; Performing entity detection, relationship extraction, and entity alignment on the multimodal cleansed data to construct a multimodal information vector; Constructing a multimodal knowledge graph based on the multimodal information vector; Based on the multimodal knowledge graph, association identification and propagation reasoning of forest pests and diseases are performed to obtain pest and disease reasoning results; Decision-making and treatment of forestry pests and diseases are carried out based on the pest and disease reasoning results.

2. The intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph according to claim 1 is characterized in that: Determining a target forestry area, obtaining multimodal forestry data of the target forestry area, and preprocessing the multimodal forestry data to obtain multimodal cleaned data specifically includes the following steps: Identify target forestry areas and multiple data types; Performing multimodal data collection on the target forestry area according to the plurality of data types to obtain multimodal forestry data; performing image enhancement and segmentation on the multimodal forestry data; Performing text cleaning on the multimodal forestry data; performing environmental data normalization on the multimodal forestry data; Generate multimodal cleaned data.

3. The intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph according to claim 1 is characterized in that: The step of performing entity detection, relationship extraction, and entity alignment on the multimodal cleansed data to construct a multimodal information vector specifically includes the following steps: Performing entity detection on the multimodal cleansed data to determine multiple target entities; Based on the plurality of target entities, performing relationship extraction on the multimodal cleansed data and recording entity relationship information; Performing multimodal entity alignment on the multimodal cleaned data and recording the entity alignment result; A multimodal information vector is constructed according to the entity relationship information, the entity alignment result and the multiple target entities.

4. The intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph according to claim 1 is characterized in that: The step of constructing a multimodal knowledge graph based on the multimodal information vector specifically includes the following steps: constructing a plurality of structured triples according to the multimodal information vector; Use graph database for management; Performing graph embedding on a plurality of the structured triples; Build a multimodal knowledge graph that integrates images, text, and environment.

5. The intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph according to claim 1 is characterized in that: The method of performing association identification and propagation reasoning of forest pests and diseases based on the multimodal knowledge graph and obtaining the pest and disease reasoning results specifically includes the following steps: Aggregating graph structure information in the multimodal knowledge graph; Performing association identification of forestry pests and diseases on the atlas structure information; Combined with the preset Bayesian path, the source and transmission path of the disease are predicted; Obtain pest and disease inference results.

6. The intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph according to claim 1 is characterized in that: The decision-making process for forestry pests and diseases based on the pest and disease reasoning results specifically includes the following steps: Based on the pest and disease reasoning results, decision-making planning for forest pests and diseases is carried out to obtain multiple pest and disease control strategies; Perform combined optimization calculations on multiple pest control strategies and select multiple target control strategies; Divide into multiple pest and disease control areas; Calculating management priority scores for the plurality of pest and disease management areas; Matching multiple corresponding governance priorities according to the multiple governance priority scores; According to the plurality of control priorities and the plurality of target control strategies, batch control of forestry pests and diseases is carried out in the plurality of pest and disease control areas.

7. The intelligent decision-making method for forestry pests and diseases based on multimodal knowledge graph according to claim 6 is characterized in that: The objective function of the combinatorial optimization calculation is: Among them, R is the total revenue, x i represents the selection state of the i-th pest control strategy, x i 1 represents selection, x i 0 means no selection, there are n pest control strategies, b i is the management benefit of the i-th pest control strategy, a i is the management cost of the i-th pest control strategy; The calculation formula for multiple governance priorities is: P j =k1H j +k2V j +k3D j ; k1+k2+k3=1; Among them, P j represents the management priority of the jth pest control area, H j represents the risk score of the jth pest control area, V j represents the forestry value of the jth pest control area, D j represents the pest density in the jth pest control area, and k1, k2, and k3 are preset weight factors.

8. The intelligent decision-making system for forestry pests and diseases based on multimodal knowledge graph is characterized by: The system includes a multimodal data acquisition unit, an entity detection and analysis unit, a knowledge graph construction unit, a forestry pest and disease reasoning unit, and a pest and disease decision processing unit, wherein: a multimodal data acquisition unit, configured to determine a target forestry area, acquire multimodal forestry data of the target forestry area, and preprocess the multimodal forestry data to acquire multimodal cleaned data; An entity detection and analysis unit, configured to perform entity detection, relationship extraction, and entity alignment processing on the multimodal cleansed data to construct a multimodal information vector; A knowledge graph construction unit, configured to construct a multimodal knowledge graph based on the multimodal information vector; A forestry pest and disease reasoning unit, configured to perform association identification and propagation reasoning of forestry pests and diseases based on the multimodal knowledge graph, and obtain pest and disease reasoning results; The pest and disease decision processing unit is used to make decisions on forest pests and diseases based on the pest and disease reasoning results.

9. The forestry pest and disease intelligent decision-making system based on multimodal knowledge graph according to claim 8 is characterized in that: The multimodal data acquisition unit specifically includes: an area determination module for determining target forestry areas and multiple data types; A multimodal data acquisition module, configured to perform multimodal data acquisition on the target forestry area according to the plurality of data types to obtain multimodal forestry data; An image enhancement module, configured to perform image enhancement and segmentation on the multimodal forestry data; A text cleaning module, used for performing text cleaning on the multimodal forestry data; A normalization processing module, used for performing environmental data normalization on the multimodal forestry data; The multimodal clean data generation module is used to generate multimodal clean data.

10. The forestry pest and disease intelligent decision-making system based on multimodal knowledge graph according to claim 8 is characterized in that: The pest and disease decision processing unit specifically includes: A decision-making and planning module is used to carry out decision-making and planning on forestry pests and diseases based on the pest and disease reasoning results, and obtain multiple pest and disease control strategies; Combinatorial optimization calculation module, used to perform combinatorial optimization calculations on multiple pest control strategies and select multiple target control strategies; Regional division module, used to divide multiple pest and disease control areas; A control priority score calculation module, used to calculate the control priority scores of the plurality of pest control areas; a priority matching module, configured to match a plurality of corresponding governance priorities according to the plurality of governance priority scores; The batch control module is used to carry out batch control of forest pests and diseases in multiple pest control areas according to multiple control priorities and multiple target control strategies.

Images