A method and system for intelligent monitoring and early warning of field blight based on multi-modal data

Abstract

This invention discloses an intelligent monitoring and early warning method and system for field diseases based on multimodal data, belonging to the field of agricultural information monitoring technology. The method divides the target field into several sub-regions using a grid, and deploys environmental sensing sensors and image acquisition devices in each sub-region. It establishes a data sampling time period and normalizes the multimodal data for each sub-region. The method calculates the environmental induction intensity value and crop characteristic deviation value, and calculates the multimodal comprehensive disease risk score. It calculates the risk evolution direction and direction sign value at adjacent time points, and calculates the induced consistency value of adjacent sub-regions. It sets a comprehensive risk threshold and an induced consistency threshold, and issues real-time early warnings for sub-regions exceeding the thresholds. This achieves closed-loop management of monitoring, analysis, and decision-making, improving the accuracy and timeliness of field disease monitoring, reducing disease spread losses, and supporting precision agricultural management and scientific prevention and control.

Description

Technical Field

[0001] This invention relates to the field of agricultural information monitoring technology, specifically to a method and system for intelligent monitoring and early warning of field diseases based on multimodal data. Background Technology

[0002] In modern agricultural production, the prevention and control of field diseases has always been a crucial aspect of ensuring crop yield and quality. With the development of information technology and the Internet of Things (IoT), digital management methods based on environmental monitoring and image recognition are increasingly being applied to field disease management, enabling early detection and dynamic tracking of diseases. In recent years, multimodal data fusion has received widespread attention in agricultural monitoring. By integrating environmental data such as temperature, humidity, leaf moisture, and wind speed, as well as characterizing information such as leaf color and mottled features, it can more comprehensively reflect crop growth status and disease risk. Simultaneously, the development of remote sensing technology and high-precision sensors has made it possible to finely divide fields and collect data in real time, providing a rich data foundation for field disease monitoring. However, existing technologies mainly rely on single data types or experience-based threshold judgment methods, which suffer from insufficient prediction accuracy, difficulty in capturing the dynamics of risk propagation, and inadequate spatial correlation analysis, making it difficult to achieve accurate monitoring and early warning of disease risks in complex field environments.

[0003] Current field disease monitoring methods still have limitations in addressing the interactions of multiple factors and the spatial spread patterns of diseases. For example, the impact of environmental triggering factors and deviations in crop physiological states on disease occurrence is difficult to quantify simultaneously in a unified model, leading to often delayed or significantly biased risk assessments. Furthermore, existing technologies lack effective means for time series analysis and identification of transmission patterns between adjacent fields, making it difficult to provide timely early warnings of abnormal transmission and limiting support for control decisions. Summary of the Invention

[0004] The purpose of this invention is to provide a field disease intelligent monitoring and early warning method and system based on multimodal data, so as to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A field disease intelligent monitoring and early warning method based on multimodal data is disclosed. The method includes the following steps: Step S1: Divide the target field's usage area into several sub-regions using a grid, generating a set of sub-regions; deploy environmental sensing sensors and image acquisition devices within each sub-region; Step S2: Establish a data sampling time period and collect multimodal data from the sub-regions; construct a monitoring dataset for the sub-regions at the data sampling time points; Step S3: Based on the monitoring dataset of the sub-regions at the data sampling time points, calculate the environmental induction intensity value and crop characteristic deviation value of the sub-regions at the data sampling time points, and calculate the multimodal comprehensive disease risk score for the sub-regions at the data sampling time points; Step S4: Calculate the risk evolution direction and direction sign of adjacent sub-regions at the data sampling time points; calculate the induced consistency value between adjacent sub-regions based on the direction sign value; Step S5: Preset the multimodal comprehensive disease risk score threshold and the induced consistency value threshold, analyze disease risk and abnormal transmission risk, and issue an early warning to relevant personnel if disease risk and abnormal transmission risk exist.

[0007] As a preferred embodiment of the intelligent monitoring and early warning method for field diseases based on multimodal data described in this invention, a high-precision remote sensing surveying device is used to survey the usage range of the target field, and the usage range of the target field is divided into several field sub-regions using a grid. These field sub-regions are then uniformly numbered, and based on the numbering information, they are classified to generate a set of field sub-regions, denoted as […]. ,in, Let A represent the a-th field sub-region, and let A represent the total number of field sub-regions.

[0008] Environmental sensing sensors and image acquisition devices are deployed in each field sub-area. The environmental sensing sensors are used to collect environmental data of the field sub-area, including temperature data, humidity data, leaf surface moisture data, and near-ground wind speed data. The image acquisition devices are used to collect crop appearance data of the field sub-area, including leaf color data and leaf mottled feature data.

[0009] As a preferred embodiment of the intelligent monitoring and early warning method for field diseases based on multimodal data described in this invention, a data sampling time period is constructed, denoted as... ,in, This represents the i-th data sampling time point, where I represents the total number of data sampling time points; at the data sampling time point Lower field area Temperature data, humidity data, leaf surface moisture data, near-ground wind speed data, leaf color data, and leaf mottled characteristic data were normalized and recorded as follows: , , , , and ;

[0010] Constructing data sampling time points Xiatian Block Area The monitoring dataset is denoted as .

[0011] As a preferred embodiment of the intelligent monitoring and early warning method for field diseases based on multimodal data described in this invention, based on data sampling time points... Xiatian Block Area Monitoring dataset Calculate the data sampling time point Xiatian Block Area The environmental induced intensity value is calculated using the following formula:

[0012] ;

[0013] in, Indicates the data sampling time point Xiatian Block Area Environmental induced intensity value;

[0014] Based on data sampling time points Xiatian Block Area Monitoring dataset Calculate the data sampling time point Xiatian Block Area The crop characterization deviation is calculated using the following formula:

[0015] ;

[0016] in, Indicates the data sampling time point Xiatian Block Area The deviation value of crop characterization.

[0017] As a preferred embodiment of the intelligent monitoring and early warning method for field diseases based on multimodal data described in this invention, based on data sampling time points... Xiatian Block Area Environmental induced intensity value and crop characterization deviation Calculate the data sampling time point Xiatian Block Area The multimodal comprehensive disease risk score is calculated using the following formula:

[0018] ;

[0019] in, Indicates the data sampling time point Xiatian Block Area Multimodal comprehensive disease risk score and These represent the preset environmental induced intensity values. and crop characterization deviation Influencing factors.

[0020] As a preferred embodiment of the intelligent monitoring and early warning method for field diseases based on multimodal data described in this invention, the data sampling time points are obtained. Xiatian Block Area Multimodal integrated disease risk score And calculate the field sub-regions at adjacent data sampling time points. The risk evolution direction quantity is denoted as ,in, ; Field sub-regions based on adjacent data sampling time points The risk evolution direction quantity is defined by the direction sign quantity, as follows:

[0021] ;

[0022] in, Indicates the data sampling time point up to the data sampling time point Xiatian Block Area Directional sign quantity;

[0023] Based on data sampling time points up to the data sampling time point Xiatian Block Area Direction sign quantity Calculate the sub-region of the field With field area The induced consistency value between them is calculated using the following formula:

[0024] ;

[0025] in, Indicates a field area With field area Induced consistency quantity between Indicates the field sub-region at adjacent data sampling time points. The direction sign quantity, Indicates the field sub-region at adjacent data sampling time points. The direction of risk evolution This represents a preset constant.

[0026] As a preferred embodiment of the intelligent monitoring and early warning method for field diseases based on multimodal data described in this invention, a multimodal comprehensive disease risk score threshold and an induced consistency threshold are preset. If the data sampling time point... Xiatian Block Area Multimodal integrated disease risk score Greater than or equal to the multimodal integrated disease risk score threshold, and the field sub-region With field area Induced consistency quantity between If the data sampling time point is greater than or equal to the induced consistency threshold, then the data sampling time point is determined. Xiatian Block Area There is a risk of disease outbreak, and at the data sampling time point up to the data sampling time point Xia Yutian Block Area If there is a risk of abnormal transmission, a warning will be issued to relevant staff.

[0027] A field disease intelligent monitoring and early warning system based on multimodal data. The system includes: a regional division and device deployment module, a data acquisition module, an index calculation and score calculation module, a symbol quantity calculation and consistency quantity calculation module, and an analysis and early warning module.

[0028] The area division and device deployment module divides the usable area of ​​the target field into several field sub-regions in a grid, generating a set of field sub-regions; and deploys environmental perception sensors and image acquisition devices in each field sub-region.

[0029] The data acquisition module: establishes a data sampling time period, collects multimodal data of field sub-regions; and constructs a monitoring dataset of the field sub-regions at the data sampling time points.

[0030] The indicator calculation and score calculation module calculates the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point based on the monitoring dataset of the field sub-region at the data sampling time point, and calculates the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point.

[0031] The symbol quantity calculation and consistency quantity calculation module calculates the risk evolution direction quantity and direction symbol quantity of field sub-regions at adjacent data sampling time points; and calculates the induced consistency quantity between adjacent field sub-regions based on the direction symbol quantity.

[0032] The analysis and early warning module: presets a multimodal comprehensive disease risk score threshold and an induced consistency threshold, analyzes disease risk and abnormal transmission risk, and issues an early warning to relevant personnel if disease risk and abnormal transmission risk exist.

[0033] Furthermore, the indicator calculation and score calculation module includes an indicator calculation unit and a score calculation unit;

[0034] The indicator calculation unit calculates the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point based on the monitoring dataset of the field sub-region at the data sampling time point.

[0035] The scoring unit calculates the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point based on the environmental induced intensity value and crop characterization deviation value of the field sub-region at the data sampling time point.

[0036] Furthermore, the symbol calculation and consistency calculation module includes a symbol calculation unit and a consistency calculation unit;

[0037] The symbol calculation unit: obtains the multimodal comprehensive disease risk score of the field sub-region at the next data sampling time point, and calculates the risk evolution direction of the field sub-region at adjacent data sampling time points; and defines the direction symbol based on the risk evolution direction of the field sub-region at adjacent data sampling time points.

[0038] The consistency calculation unit calculates the induced consistency between adjacent field sub-regions based on the directional sign value of the field sub-regions at adjacent data sampling time points.

[0039] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: This invention provides a field disease intelligent monitoring and early warning method and system based on multimodal data. By conducting high-precision remote sensing surveys of target fields and dividing them into several sub-regions using a grid, and deploying environmental sensing sensors and image acquisition devices in each sub-region, comprehensive collection of multimodal data representing the environment and crops is achieved, providing a high-resolution spatial foundation for subsequent data analysis. Within the constructed data sampling time period, multimodal data from each sub-region is periodically collected and normalized to form a standardized monitoring dataset, making data from different time points comparable and traceable. Based on the monitoring dataset, environmental induction factors are calculated. By analyzing the intensity value and crop characteristic deviation value, and calculating the multimodal comprehensive disease risk score through the fusion of preset influencing factors, a quantitative assessment of the potential disease risk in sub-regions of the field was achieved. Furthermore, by calculating the risk evolution direction and direction sign at adjacent time points, and combining the induced consistency value of adjacent sub-regions, dynamic analysis of disease risk in time and space and identification of potential transmission trends were realized. Finally, by setting the comprehensive risk threshold and induced consistency threshold, real-time early warnings were issued for sub-regions exceeding the standard, realizing closed-loop management from monitoring, analysis to decision-making. This significantly improved the accuracy, timeliness, and scientific prevention and control capabilities of field disease monitoring, reduced disease transmission losses, and supported precision agricultural management. Attached Figure Description

[0040] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0041] Figure 1 This is a schematic diagram illustrating the steps of an intelligent monitoring and early warning method for field diseases based on multimodal data according to the present invention.

[0042] Figure 2 This is a schematic diagram of the structure of a field disease intelligent monitoring and early warning system based on multimodal data according to the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Please see Figure 1 In this first embodiment, a field disease intelligent monitoring and early warning method based on multimodal data is provided. The method includes the following steps:

[0045] Step S1: Divide the usage area of ​​the target field into several sub-regions by gridding, and generate a set of sub-regions; deploy environmental perception sensors and image acquisition devices in each sub-region.

[0046] Specifically, using high-precision remote sensing equipment, the usage area of ​​the target field is surveyed, and the usage area of ​​the target field is divided into several sub-regions using a grid. These sub-regions are then uniformly numbered, classified based on their numbering information, and a set of sub-regions is generated, denoted as […]. ,in, Let A represent the a-th field sub-region, and let A represent the total number of field sub-regions.

[0047] Environmental sensing sensors and image acquisition devices are deployed in each field sub-area. The environmental sensing sensors are used to collect environmental data of the field sub-area, including temperature data, humidity data, leaf surface moisture data, and near-ground wind speed data. The image acquisition devices are used to collect crop appearance data of the field sub-area, including leaf color data and leaf mottled feature data.

[0048] In this invention, high-precision remote sensing equipment is used to accurately survey the target field, and the field is divided into several sub-regions by gridding. This enables systematic management and numbering of the field's spatial information, providing a clear spatial index and management structure for subsequent data collection. Environmental sensing sensors and image acquisition devices are deployed within each sub-region, enabling the simultaneous acquisition of both environmental and crop characterization information, ensuring the integrity and consistency of multimodal data.

[0049] This step, through refined spatial division and multimodal sensor deployment, enables subsequent real-time monitoring of fields to have high spatial resolution, comprehensive data, and strong traceability, thereby improving the accuracy and quantifiability of field disease risk monitoring and laying the foundation for scientific decision-making and precise early warning.

[0050] Step S2: Construct a data sampling time period and collect multimodal data of field sub-regions; construct a monitoring dataset of field sub-regions at the data sampling time points.

[0051] Specifically, the data sampling time period is constructed, denoted as . ,in, This represents the i-th data sampling time point, where I represents the total number of data sampling time points; at the data sampling time point Lower field area Temperature data, humidity data, leaf surface moisture data, near-ground wind speed data, leaf color data, and leaf mottled characteristic data were normalized and recorded as follows: , , , , and ;

[0052] Constructing data sampling time points Xiatian Block Area The monitoring dataset is denoted as .

[0053] In this invention, continuous monitoring over time is achieved by constructing a data sampling time period and collecting multimodal data such as temperature, humidity, leaf surface moisture, near-ground wind speed, leaf color, and leaf mottle characteristics at each time point. Simultaneously, the various types of data are normalized to form a monitoring dataset, providing standardized input for unified data analysis.

[0054] It enables the systematic and standardized storage of multimodal data of field sub-regions, allowing direct comparative analysis of data at different time points, supporting dynamic disease risk calculation, improving data comparability and processing efficiency, and providing a reliable basis for subsequent calculation of environmental induced intensity and crop deviation, thereby enhancing the accuracy of disease risk assessment.

[0055] Step S3: Based on the monitoring dataset of the field sub-region at the data sampling time point, calculate the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point, and calculate the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point.

[0056] Specifically, based on the data sampling time point Xiatian Block Area Monitoring dataset Calculate the data sampling time point Xiatian Block Area The environmental induced intensity value is calculated using the following formula:

[0057] ;

[0058] in, Indicates the data sampling time point Xiatian Block Area Environmental induced intensity value;

[0059] It should be noted that temperature × humidity reflects the basic climate suitability, and the square of leaf surface moisture amplifies its direct inducing effect. In actual field practice, the duration of leaf surface moisture exceeding 4 hours is the key threshold for the outbreak of fungal diseases (such as powdery mildew and downy mildew), and the square term can amplify the influence of this key factor. The greater the wind speed, the more difficult it is for fungal spores to attach to the crop surface. Therefore, "1 + wind speed" is used as a correction term. The greater the wind speed, the lower the environmental induction intensity value, which is in line with the physical laws of fungal transmission in the field.

[0060] This formula quantifies whether environmental conditions support the occurrence of disease: the higher the value, the more likely the climate and leaf condition of the sub-region are to induce disease, providing basic environmental data for subsequent risk assessment.

[0061] in the formula By amplifying the influence weight of key inducing factors, in actual field practice, the duration of leaf surface moisture is a decisive condition for fungal spore germination (e.g., downy mildew requires leaf surface moisture for ≥6 hours). The squared term can quickly reflect abnormal changes in this factor (e.g., a sudden increase in humidity after rainfall) in the indicators, avoiding key signals being masked by other factors.

[0062] Based on data sampling time points Xiatian Block Area Monitoring dataset Calculate the data sampling time point Xiatian Block Area The crop characterization deviation is calculated using the following formula:

[0063] ;

[0064] in, Indicates the data sampling time point Xiatian Block Area The deviation of crop characterization;

[0065] It should be noted that leaf color and mottled features are considered as two coordinates in a two-dimensional space. The deviation value represents the distance between the actual state and the healthy state (normalized baseline value). This approach amplifies significant deviations of a single feature (such as severe mottled) while also taking into account synergistic deviations of two features (such as chlorosis + dense spots), aligning with the manifestation patterns of disease symptoms in the field. Higher values ​​indicate greater deviations from the crop's health status and a higher probability of disease occurrence, providing direct crop-level evidence for risk assessment.

[0066] Furthermore, based on the data sampling time point Xiatian Block Area Environmental induced intensity value and crop characterization deviation Calculate the data sampling time point Xiatian Block Area The multimodal comprehensive disease risk score is calculated using the following formula:

[0067] ;

[0068] in, Indicates the data sampling time point Xiatian Block Area Multimodal comprehensive disease risk score and These represent the preset environmental induced intensity values. and crop characterization deviation Influencing factors.

[0069] It should be noted that the combined effect of environmental potential and crop abnormality means that, in actual field practice, disease outbreaks are often the result of a suitable environment plus crop susceptibility. By integrating the two heterogeneous data of environment and crop into quantifiable and comparable risk indicators, the core problem of data fragmentation and inability to make comprehensive judgments in traditional methods is solved.

[0070] In this invention, the environmental induction intensity is calculated by weighted combination of temperature, humidity, leaf surface moisture, and near-ground wind speed, enabling a quantitative assessment of the potential of environmental conditions for disease occurrence. Crop deviation values ​​are calculated based on leaf color and mottled characteristics, providing a quantitative description of deviations in crop health status. By integrating environmental and crop deviations according to preset influencing factors to calculate comprehensive risk, the quantification of disease risk under the synergistic effect of multiple factors is achieved.

[0071] This step, through the quantification and fusion of multiple indicators, achieves a comprehensive assessment of disease risks that cannot be reflected by a single indicator. It enables early identification of potential disease threats in sub-field areas, providing a more accurate and scientific disease risk assessment, making early warning more forward-looking and targeted, and reducing the misjudgment that may be caused by a single data indicator.

[0072] Step S4: Calculate the risk evolution direction and direction sign of the sub-region of the field at adjacent data sampling time points; based on the direction sign, calculate the induced consistency between adjacent sub-regions of the field.

[0073] Specifically, obtain the data sampling time point. Xiatian Block Area Multimodal integrated disease risk score And calculate the field sub-regions at adjacent data sampling time points. The risk evolution direction quantity is denoted as ,in, ; Field sub-regions based on adjacent data sampling time points The risk evolution direction quantity is defined by the direction sign quantity, as follows:

[0074] ;

[0075] in, Indicates the data sampling time point up to the data sampling time point Xiatian Block Area Directional sign quantity;

[0076] Based on data sampling time points up to the data sampling time point Xiatian Block Area Direction sign quantity Calculate the sub-region of the field With field area The induced consistency value between them is calculated using the following formula:

[0077] ;

[0078] in, Indicates a field area With field area Induced consistency quantity between Indicates the field sub-region at adjacent data sampling time points. The direction sign quantity, Indicates the field sub-region at adjacent data sampling time points. The direction of risk evolution This represents a preset constant.

[0079] It should be noted that, if A value of 1 indicates that the risk trends in the two sub-regions are the same (e.g., both are rising), and there is a propagation correlation; a value of -1 indicates that the trends are opposite, and the propagation correlation is weak; a value of 0 indicates that one side has no change, and the correlation is low. The ratio reflects the consistency in the intensity of risk changes between the two sub-regions; the closer the ratio is to 1, the stronger the correlation. The magnitude of risk changes and The closer they are, the higher the likelihood of transmission;

[0080] In this invention, by calculating the risk evolution direction and direction sign at adjacent time points, the dynamic tracking of the disease risk trend over time is achieved. Based on the direction sign, the induced consistency of adjacent field sub-regions is calculated, enabling a quantitative assessment of the disease transmission trend in the spatial dimension.

[0081] It enables dynamic analysis of disease risk in time and space, which can not only identify risk levels, but also reveal potential abnormal transmission paths.

[0082] This step, through dynamic trend analysis and spatially induced consistency calculation, ensures that early warning is not only based on single-point risk levels but also considers risk transmission relationships, thereby improving the accuracy of identifying abnormal disease transmission and enhancing the scientific nature of prevention and control decisions.

[0083] Step S5: Preset the multimodal comprehensive disease risk score threshold and the induced consistency threshold, analyze the disease risk and abnormal transmission risk, and issue an early warning to relevant staff if disease risk and abnormal transmission risk exist.

[0084] Specifically, preset thresholds for multimodal comprehensive disease risk score and induced consistency threshold, if the data sampling time point Xiatian Block Area Multimodal integrated disease risk score Greater than or equal to the multimodal integrated disease risk score threshold, and the field sub-region With field area Induced consistency quantity between If the data sampling time point is greater than or equal to the induced consistency threshold, then the data sampling time point is determined. Xiatian Block Area There is a risk of disease outbreak, and at the data sampling time point up to the data sampling time point Xia Yutian Block Area If there is a risk of abnormal transmission, a warning will be issued to relevant staff.

[0085] Please see Figure 2In this second embodiment: a field disease intelligent monitoring and early warning system based on multimodal data is provided. The system includes: a regional division and device deployment module, a data acquisition module, an index calculation and score calculation module, a symbol quantity calculation and consistency quantity calculation module, and an analysis and early warning module.

[0086] The area division and device deployment module divides the usable area of ​​the target field into several field sub-regions in a grid, generating a set of field sub-regions; and deploys environmental perception sensors and image acquisition devices in each field sub-region.

[0087] The data acquisition module: establishes a data sampling time period, collects multimodal data of field sub-regions; and constructs a monitoring dataset of the field sub-regions at the data sampling time points.

[0088] The indicator calculation and score calculation module calculates the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point based on the monitoring dataset of the field sub-region at the data sampling time point, and calculates the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point.

[0089] The symbol quantity calculation and consistency quantity calculation module calculates the risk evolution direction quantity and direction symbol quantity of field sub-regions at adjacent data sampling time points; and calculates the induced consistency quantity between adjacent field sub-regions based on the direction symbol quantity.

[0090] The analysis and early warning module: presets a multimodal comprehensive disease risk score threshold and an induced consistency threshold, analyzes disease risk and abnormal transmission risk, and issues an early warning to relevant personnel if disease risk and abnormal transmission risk exist.

[0091] Furthermore, the indicator calculation and score calculation module includes an indicator calculation unit and a score calculation unit;

[0092] The indicator calculation unit calculates the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point based on the monitoring dataset of the field sub-region at the data sampling time point.

[0093] The scoring unit calculates the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point based on the environmental induced intensity value and crop characterization deviation value of the field sub-region at the data sampling time point.

[0094] Furthermore, the symbol calculation and consistency calculation module includes a symbol calculation unit and a consistency calculation unit;

[0095] The symbol calculation unit: obtains the multimodal comprehensive disease risk score of the field sub-region at the next data sampling time point, and calculates the risk evolution direction of the field sub-region at adjacent data sampling time points; and defines the direction symbol based on the risk evolution direction of the field sub-region at adjacent data sampling time points.

[0096] The consistency calculation unit calculates the induced consistency between adjacent field sub-regions based on the directional sign value of the field sub-regions at adjacent data sampling time points.

[0097] In this third embodiment, a field blight intelligent monitoring and early warning method based on multimodal data is provided. Taking rice blast disease monitoring in rice fields as an example, the application process of the method of the present invention in rice production scenarios is specifically illustrated:

[0098] A 10-mu (approximately 1.65 acres) paddy field was selected as the target field. A high-precision UAV equipped with multispectral remote sensing equipment was used for surveying. The field was divided into 100 sub-regions using a 10m × 10m grid, and each sub-region was uniformly numbered. , ,…, Environmental sensing sensors (for collecting data on temperature, humidity, leaf moisture, and near-ground wind speed) and image acquisition devices (for collecting crop appearance data, i.e., leaf color data and leaf mottled feature data) are deployed in each sub-area.

[0099] The sampling period is set to twice daily, at 8:00 AM and 2:00 PM. This represents the i-th sampling time point. Environmental data (temperature data, humidity data, leaf surface moisture data, near-ground wind speed data) of each sub-region were collected, along with crop appearance data obtained through image acquisition devices. Leaf color data and leaf mottle feature data were extracted after image analysis. All data were normalized to construct the monitoring dataset.

[0100] ;

[0101] as subregion For example, the normalized data from a sample collected at 8:00 AM on a certain day is as follows: , , , , , Calculate the environmental induced intensity value:

[0102] ;

[0103] It should be noted that the product of temperature and humidity reflects the promoting effect of hot and humid environments on pathogen reproduction. Rice blast fungus exhibits the highest activity at 25–28℃ and humidity >90%, and this term captures this synergistic effect. Sustained leaf moisture is a key condition for fungal spore germination; the square term amplifies its impact, causing this index to rise rapidly after continuous rain or dew, providing an early warning of disease risk. Higher wind speeds are less conducive to spore attachment and disease spread; therefore, its introduction as an inhibitor aligns with the physical laws governing disease spread in the field.

[0104] This formula quantifies the potential of field microclimate to induce airborne or waterborne diseases such as rice blast and sheath blight, providing a numerical basis for environmental risk assessment.

[0105] Calculate crop characterization deviation:

[0106] ;

[0107] It should be noted that leaf color reflects chlorophyll content and photosynthetic capacity. Color deviations (such as yellowing or fading) often indicate nutrient deficiency or early symptoms of disease. Leaf mottling directly corresponds to the area and morphology of lesions, providing direct visual evidence of disease occurrence. By treating color and mottling as coordinates in a two-dimensional health state space and calculating their geometric distance from the ideal health state (coordinates), the degree of apparent abnormality in the crop canopy can be comprehensively reflected. This formula achieves a quantitative description of the health status of the rice canopy, and is particularly suitable for identifying diseases with typical leaf symptoms such as rice blast, brown spot, and bacterial blight.

[0108] Let the environmental and crop influencing factors be respectively , The multimodal comprehensive disease risk score is:

[0109] ;

[0110] At 2:00 PM, the sub-region was measured. The risk score rose to Adjacent sub-regions Score at 8:00 AM Score at 2:00 PM Calculate the risk evolution direction:

[0111] ,but ;

[0112] ,but ;

[0113] Calculate the induced consistency quantity (assuming...) ):

[0114] ;

[0115] The preset threshold for the multimodal comprehensive disease risk score is 0.5, and the threshold for induced consistency is 1.5. Because... and Then the system determines the sub-region There is a significant risk of rice blast, and it is adjacent to the neighboring sub-region. If an abnormal spread trend is detected, an early warning message will be immediately pushed to farmers' mobile apps, and it will be recommended to implement targeted spraying and drainage management in the area.

[0116] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0117] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring and early warning of field diseases based on multi-modal data, characterized in that, The method includes the following steps: Step S1: Divide the target field's usage area into several sub-regions using a grid, generating a set of sub-regions; Deploy environmental sensing sensors and image acquisition devices within each sub-region; Step S2: Establish the data sampling time period and collect multimodal data of field sub-regions; construct the monitoring dataset of field sub-regions at the data sampling time points; Step S3: Based on the monitoring dataset of the field sub-region at the data sampling time point, calculate the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point, and calculate the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point; Step S4: Calculate the risk evolution direction and direction sign of the sub-regions of the field at adjacent data sampling time points; based on the direction sign, calculate the induced consistency between adjacent sub-regions of the field. Step S5: Preset the multimodal comprehensive disease risk score threshold and the induced consistency threshold, analyze the disease risk and abnormal transmission risk, and issue an early warning to relevant staff if disease risk and abnormal transmission risk exist.

2. The method of claim 1, wherein the method comprises: The specific implementation process of step S1 includes: Using high-precision remote sensing equipment, the usage area of ​​the target field is surveyed, and the usage area of ​​the target field is divided into several sub-regions using a grid. These sub-regions are then uniformly numbered, classified based on their numbering information, and a set of sub-regions is generated, denoted as […]. ,in, Let A represent the a-th field sub-region, and let A represent the total number of field sub-regions. Environmental sensing sensors and image acquisition devices are deployed in each field sub-area. The environmental sensing sensors are used to collect environmental data of the field sub-area, including temperature data, humidity data, leaf surface moisture data, and near-ground wind speed data. The image acquisition devices are used to collect crop appearance data of the field sub-area, including leaf color data and leaf mottled feature data.

3. The intelligent monitoring and early warning method for field diseases based on multimodal data according to claim 2, characterized in that, The specific implementation process of step S2 includes: The data sampling time period is defined as follows: ,in, This represents the i-th data sampling time point, where I represents the total number of data sampling time points; at the data sampling time point Lower field area Temperature data, humidity data, leaf surface moisture data, near-ground wind speed data, leaf color data, and leaf mottled characteristic data were normalized and recorded as follows: , , , , and ; Constructing data sampling time points Xiatian Block Area The monitoring dataset is denoted as .

4. The intelligent monitoring and early warning method for field diseases based on multimodal data according to claim 3, characterized in that, The specific implementation process of step S3 includes: Based on data sampling time points Xiatian Block Area Monitoring dataset Calculate the data sampling time point Xiatian Block Area The environmental induced intensity value is calculated using the following formula: ； in, Indicates the data sampling time point Xiatian Block Area Environmental induced intensity value; Based on data sampling time points Xiatian Block Area Monitoring dataset Calculate the data sampling time point Xiatian Block Area The crop characterization deviation is calculated using the following formula: ； in, Indicates the data sampling time point Xiatian Block Area The deviation value of crop characterization.

5. The intelligent monitoring and early warning method for field diseases based on multimodal data according to claim 4, characterized in that, The specific implementation process of step S3 also includes: Based on data sampling time points Xiatian Block Area Environmental induced intensity value and crop characterization deviation Calculate the data sampling time point Xiatian Block Area The multimodal comprehensive disease risk score is calculated using the following formula: ； in, Indicates the data sampling time point Xiatian Block Area Multimodal comprehensive disease risk score and These represent the preset environmental induced intensity values. and crop characterization deviation Influencing factors.

6. The intelligent monitoring and early warning method for field diseases based on multimodal data according to claim 5, characterized in that, The specific implementation process of step S4 includes: Obtain data sampling time points Xiatian Block Area Multimodal integrated disease risk score And calculate the field sub-regions at adjacent data sampling time points. The risk evolution direction quantity is denoted as ,in, ; Field sub-regions based on adjacent data sampling time points The risk evolution direction quantity is defined by the direction sign quantity, as follows: ； in, Indicates the data sampling time point up to the data sampling time point Xiatian Block Area Directional sign quantity; Based on data sampling time points up to the data sampling time point Xiatian Block Area Direction sign quantity Calculate the sub-region of the field With field area The induced consistency value between them is calculated using the following formula: ； in, Indicates a field area With field area Induced consistency quantity between Indicates the field sub-region at adjacent data sampling time points. The direction sign quantity, Indicates the field sub-region at adjacent data sampling time points. The direction of risk evolution This represents a preset constant.

7. The intelligent monitoring and early warning method for field diseases based on multimodal data according to claim 6, characterized in that, The specific implementation process of step S5 includes: Preset multimodal comprehensive disease risk score threshold and induced consistency threshold, if the data sampling time point Xiatian Block Area Multimodal integrated disease risk score Greater than or equal to the multimodal integrated disease risk score threshold, and the field sub-region With field area Induced consistency quantity between If the data sampling time point is greater than or equal to the induced consistency threshold, then the data sampling time point is determined. Xiatian Block Area There is a risk of disease outbreak, and at the data sampling time point up to the data sampling time point Xia Yutian Block Area If there is a risk of abnormal transmission, a warning will be issued to relevant staff.

8. A field disease intelligent monitoring and early warning system based on multimodal data, executing the field disease intelligent monitoring and early warning method based on multimodal data as described in any one of claims 1-7, characterized in that, The system includes: a region division and device deployment module, a data acquisition module, an indicator calculation and score calculation module, a symbol quantity calculation and consistency quantity calculation module, and an analysis and early warning module; The area division and device deployment module divides the usable area of ​​the target field into several field sub-regions in a grid, generating a set of field sub-regions; and deploys environmental perception sensors and image acquisition devices in each field sub-region. The data acquisition module: establishes a data sampling time period, collects multimodal data of field sub-regions; and constructs a monitoring dataset of the field sub-regions at the data sampling time points. The indicator calculation and score calculation module calculates the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point based on the monitoring dataset of the field sub-region at the data sampling time point, and calculates the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point. The symbol quantity calculation and consistency quantity calculation module calculates the risk evolution direction quantity and direction symbol quantity of field sub-regions at adjacent data sampling time points; and calculates the induced consistency quantity between adjacent field sub-regions based on the direction symbol quantity. The analysis and early warning module: presets a multimodal comprehensive disease risk score threshold and an induced consistency threshold, analyzes disease risk and abnormal transmission risk, and issues an early warning to relevant personnel if disease risk and abnormal transmission risk exist.

9. A field disease intelligent monitoring and early warning system based on multimodal data according to claim 8, characterized in that: The indicator calculation and score calculation module includes an indicator calculation unit and a score calculation unit; The indicator calculation unit calculates the environmental induction intensity value and crop characterization deviation value of the field sub-region at the data sampling time point based on the monitoring dataset of the field sub-region at the data sampling time point. The scoring unit calculates the multimodal comprehensive disease risk score of the field sub-region at the data sampling time point based on the environmental induced intensity value and crop characterization deviation value of the field sub-region at the data sampling time point.

10. A field disease intelligent monitoring and early warning system based on multimodal data according to claim 9, characterized in that: The symbol calculation and consistency calculation module includes a symbol calculation unit and a consistency calculation unit; The symbol calculation unit: obtains the multimodal comprehensive disease risk score of the field sub-region at the next data sampling time point, and calculates the risk evolution direction of the field sub-region at adjacent data sampling time points; and defines the direction symbol based on the risk evolution direction of the field sub-region at adjacent data sampling time points. The consistency calculation unit calculates the induced consistency between adjacent field sub-regions based on the directional sign value of the field sub-regions at adjacent data sampling time points.

Images