A method and system for identifying viral diseases in cucurbit vegetables based on sensor arrays

Abstract

This invention discloses a method and system for identifying viral diseases in cucurbit vegetables based on a sensor array. The method generates a dynamic heat map by collecting insect activity data from the target planting area, and constructs a viral disease transmission distribution map based on an insect vector transmission probability model, thereby deploying an initial sensor array. It then acquires real-time plant physiological parameters and environmental data through multiple sensors to construct a viral disease identification model that integrates insect vector activity characteristics, enabling early disease diagnosis. Based on the identification results, the sensor array performance is evaluated, and an adaptive optimization algorithm is used to dynamically adjust the array layout, forming an optimized network that balances monitoring accuracy and resource efficiency. This significantly improves the sensitivity and accuracy of viral disease detection, reduces monitoring costs, and provides a digital solution for the prevention and control of diseases in cucurbit crops.

Description

Technical Field

[0001] This invention relates to the field of cucurbit vegetable disease identification technology, and in particular to a method and system for identifying viral diseases in cucurbit vegetables based on a sensor array. Background Technology

[0002] With the advancement of agricultural modernization, especially in the cultivation of cucurbits and vegetables, the prevention and control of viral diseases has increasingly become a key factor affecting crop yield and quality. Traditional methods of viral disease monitoring and control mainly rely on manual inspections and conventional detection techniques, which cannot effectively guarantee the accuracy and timeliness of the test results. Therefore, developing an efficient, accurate, and automated viral disease identification technology is of great significance for improving agricultural production efficiency and reducing losses.

[0003] Cucurbit vegetables are frequently attacked by various viral diseases during cultivation, severely impacting crop yield and quality. Studies have shown that insect-borne vectors, particularly aphids and whiteflies, are among the main vectors for viral disease transmission. These insects can carry viruses from diseased plants to healthy plants during feeding, leading to rapid spread of the virus in the field. Traditional methods for identifying viral diseases rely heavily on manual inspections and laboratory testing, which are inefficient, inaccurate, and fail to provide timely updates on disease progression.

[0004] In recent years, with the development of intelligent agriculture, disease monitoring technology based on sensor arrays has been gradually applied to field management. By deploying spectral sensors and high-resolution camera equipment, crop phenotypic characteristics and growth status can be obtained, providing data support for disease identification. However, existing technologies generally neglect the spatial characteristics and dynamic patterns of viral and insect vector transmission, sensor deployment lacks specificity, and identification accuracy and coverage efficiency need to be improved.

[0005] Therefore, developing a viral disease identification method that integrates insect vector activity information, transmission probability modeling, and sensor array optimization has become a key direction for precision agriculture prevention and control. This invention addresses these issues by proposing a method and system for identifying viral diseases in cucurbit vegetables that integrates insect activity heatmaps, transmission probability distributions, and intelligent identification models, significantly improving the timeliness and accuracy of identification. Summary of the Invention

[0006] To address at least one of the aforementioned technical problems, this invention proposes a method and system for identifying viral diseases in cucurbit vegetables based on a sensor array.

[0007] The first aspect of this invention provides a method for identifying viral diseases in cucurbit vegetables based on a sensor array, comprising:

[0008] Data on insect activity in the target planting area of ​​cucurbit vegetables is obtained, and a heat map of insect activity in the target planting area is constructed based on the insect activity data;

[0009] Based on the insect activity heat map, construct an insect vector transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area, and construct an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the insect vector transmission probability distribution map;

[0010] The virus disease monitoring data of each sensor is acquired, a virus disease identification model for cucurbit vegetables is constructed, and the virus disease monitoring data is identified according to the virus disease identification model to obtain the virus disease identification results for cucurbit vegetables;

[0011] The virus identification quality of the initial sensor array is determined based on the identification results, and the initial sensor array is adjusted based on the virus identification quality to obtain an optimized sensor array.

[0012] In this solution, the step of acquiring insect activity data in the target planting area of ​​cucurbit vegetables and constructing an insect activity heat map of the target planting area based on the insect activity data specifically involves:

[0013] Historical insect infestation records of the target planting area for cucurbit vegetables are obtained. Based on the historical insect infestation records, the types of active insects in the target planting area are determined. Standard image data of the active insect types are obtained. Insect names are labeled on the standard image data to obtain labeled image data.

[0014] An insect recognition model is constructed based on a convolutional neural network. The model consists of convolutional layers, pooling layers, and fully connected layers. A cross-entropy loss function and an Adam optimizer are also constructed. The labeled image data is then imported into the insect recognition model for training.

[0015] The target planting area is divided into N sub-regions of a preset size. Image data of each sub-region is acquired within a preset time period using a high-definition camera. The image data is then imported into the insect recognition model for insect recognition to obtain insect activity data for each sub-region. The insect activity data includes insect type and insect quantity.

[0016] Based on the kernel density estimation algorithm, each sub-region is regarded as an observation point, the number of insects is used as the weight value of the observation point, a Gaussian kernel function is selected and the bandwidth parameter of the Gaussian kernel function is determined, and a continuous spatial insect density distribution map is constructed in the target planting area according to the kernel density estimation algorithm.

[0017] Based on the insect density distribution map, the insect density at each location is color-mapped to construct a heat map of insect activity in the target planting area over a preset time period.

[0018] In this scheme, the step of constructing a vector-borne transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area based on the insect activity heat map, and constructing an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the vector-borne transmission probability distribution map, specifically involves:

[0019] Data on viral disease incidence in cucurbit vegetables in the target planting area during the preset time period are obtained. The data includes the location of the disease, the type of viral disease, the severity of the disease, and the spread path of the disease.

[0020] Spatial overlay analysis was performed on the insect types, insect densities, and disease severity in the corresponding sub-regions of the insect activity heatmaps. Pearson correlation coefficients between the density of each insect type and the severity of disease were calculated, and a correlation coefficient matrix was constructed. Based on the correlation coefficient matrix, target insect types with a correlation coefficient greater than a preset value for viral disease transmission were selected.

[0021] Based on the insect activity heatmap, the distribution density data of the target insect type in the target planting area is extracted. Combined with the directional characteristics of the spread path of the disease location, a virus transmission and diffusion model based on the migration path of the target insect is constructed.

[0022] The virus transmission probability weight of each sub-region is calculated based on the virus transmission and diffusion model. Sub-regions that are closer to the onset of disease and have a higher density of target insects are given a higher transmission probability weight.

[0023] Spatial interpolation is performed on the propagation probability weights of each sub-region to generate a continuous probability surface covering the target planting area. Based on the probability surface threshold, high, medium and low propagation risk areas are divided to construct an insect vector propagation probability distribution map.

[0024] The density of hyperspectral imaging sensors in each sub-region of the target planting area is determined based on the vector propagation probability distribution map. The placement location is determined based on the density. An initial sensor array for identifying viral diseases in cucurbit vegetables in the target planting area is constructed based on the placement location.

[0025] In this scheme, the acquisition of viral disease monitoring data from each sensor, the construction of a viral disease identification model for cucurbit vegetables, and the identification of the viral disease monitoring data based on the viral disease identification model to obtain the viral disease identification result for cucurbit vegetables are specifically as follows:

[0026] The appearance morphology data of plants with different types of viral diseases at different growth stages are obtained, and the spectral reflectance data of the appearance morphology are obtained. The spectral features of the spectral reflectance data are extracted based on the PCA algorithm. The spectral features of the same type of viral disease are integrated to construct the spectral feature vector of plants with viral diseases.

[0027] The spectral feature vectors are labeled with the viral disease type and severity to obtain training set labeled spectral feature data;

[0028] A virus disease classification model is constructed based on the support vector machine algorithm. A Gaussian kernel function is selected and the Hierarchy loss function and L2 regularization term are defined. The sequence minimum optimization algorithm is used to train the model on the label spectral feature data of the training set. The kernel function parameters and regularization coefficients are adjusted through cross-validation until the classification accuracy exceeds the preset threshold.

[0029] The trained viral disease classification model is deployed to each sensor node in the initial sensor array. The spectral reflectance data monitored by the sensors is acquired in real time and spectral features are extracted. The extracted spectral features are input into the viral disease classification model to predict the type and severity level of the viral disease. The prediction results of all sensors are summarized and mapped to the corresponding sub-regions to generate the viral disease identification results of the target planting area.

[0030] In this scheme, determining the virus identification quality of the initial sensor array based on the identification results, and adjusting the initial sensor array based on the virus identification quality to obtain an optimized sensor array, specifically involves:

[0031] The actual viral disease infection data of cucurbit vegetables in each sub-region of the target planting area is obtained. The actual viral disease infection data is compared with the viral disease identification results to determine the accuracy and false negative rate of the initial sensor array for viral disease identification in each sub-region.

[0032] The identification quality of cucurbit vegetables in each sub-region was evaluated based on the accuracy and false negative rate of the virus disease identification, and the initial sensor array's virus disease identification quality data for each sub-region was obtained.

[0033] The virus disease identification quality data is clustered based on the BI RCH clustering algorithm, dividing the target planting area into high identification quality area, medium identification quality area and low identification quality area, and generating a spatially distributed identification quality cluster map.

[0034] The identification quality clustering map and the vector transmission probability distribution map are spatially overlaid to extract the proportion of the area covered by the high identification quality area in the high-risk area of ​​vector transmission, and the identification coverage rate of the vector transmission risk area is calculated.

[0035] If the coverage rate of the vector-borne transmission risk area identification is less than a preset threshold, then sensor deployment defect analysis is performed on the uncovered high-risk vector-borne transmission sub-regions to obtain the mean value of virus transmission probability weight and the corresponding virus disease identification quality score of the defective sub-regions.

[0036] When the average virus propagation probability weight of the defective sub-region is greater than the preset risk threshold and the virus disease identification quality score is lower than the preset quality threshold, it is determined to be a sensor deployment failure area. The ant colony optimization algorithm is used to re-plan the sensor deployment path in the failure area, and the density of hyperspectral imaging sensors is increased at the peak point of the virus propagation probability weight to obtain the first optimization strategy.

[0037] When the average weight of the virus transmission probability in the defective sub-region is less than the preset risk threshold but the identification quality score is lower than the preset quality threshold, it is determined to be a sensor monitoring blind zone. The monitoring angle and spectral acquisition frequency of the sensor in the blind zone are adjusted according to the migration direction of the target insect in the insect activity heat map to obtain the second optimization strategy.

[0038] The initial sensor array is adjusted according to the first optimization strategy and the second optimization strategy to obtain an optimized sensor array.

[0039] In this scheme, the clustering operation on the virus identification quality data based on the BI RCH clustering algorithm specifically involves:

[0040] The viral disease identification quality data is subjected to data standardization processing, and the feature values ​​of each dimension are transformed into zero mean and unit variance distributions to obtain standardized viral disease data.

[0041] The BI RCH clustering algorithm is introduced to calculate the branch factor threshold based on the data dimension and sample number of the standardized viral disease data, and to determine the maximum branch factor and sub-cluster diameter threshold of the clustering feature tree.

[0042] The clustering feature tree structure is initialized based on the maximum branching factor and the sub-cluster diameter threshold, and a root node containing an empty clustering feature vector is created. The clustering feature vector consists of the number of samples, a linear summation vector, and a sum of squares scalar.

[0043] Traverse each data point in the standardized viral disease data, starting from the root node and searching downwards along the tree structure. Calculate the Euclidean distance between the current data point and the cluster feature vectors of each sub-cluster, and select the sub-cluster path with the closest distance for recursive access.

[0044] If the radius of a subcluster in a leaf node does not exceed the subcluster diameter threshold, the current data point is merged into the subcluster and the clustering feature vector is updated; otherwise, a new subcluster is created and the branching factor limit is checked. If the branching factor limit is exceeded, the node is split to generate a new leaf node.

[0045] After all data points have been inserted, the clustering feature tree is condensed, and the subclusters in adjacent leaf nodes that meet the diameter threshold are merged to form coarse-grained clusters.

[0046] Based on the coarse-grained clustering results, the clustering feature vectors of all sub-clusters are extracted as a new dataset. A hierarchical clustering algorithm is used to perform global clustering operations. The distance matrix between sub-clusters is calculated and the final number of clusters is determined by cutting the dendrogram.

[0047] The viral disease identification quality data is mapped to the global clustering results according to the sub-cluster affiliation relationship, thus obtaining the clustering results of the viral disease identification quality data.

[0048] A second aspect of the present invention also provides a system for identifying viral diseases of cucurbit vegetables based on a sensor array. The system includes a memory and a processor. The memory includes a program for identifying viral diseases of cucurbit vegetables based on a sensor array. When the processor executes the program for identifying viral diseases of cucurbit vegetables based on a sensor array, it performs the following steps:

[0049] Data on insect activity in the target planting area of ​​cucurbit vegetables is obtained, and a heat map of insect activity in the target planting area is constructed based on the insect activity data;

[0050] Based on the insect activity heat map, construct an insect vector transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area, and construct an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the insect vector transmission probability distribution map;

[0051] The virus disease monitoring data of each sensor is acquired, a virus disease identification model for cucurbit vegetables is constructed, and the virus disease monitoring data is identified according to the virus disease identification model to obtain the virus disease identification results for cucurbit vegetables;

[0052] The virus identification quality of the initial sensor array is determined based on the identification results, and the initial sensor array is adjusted based on the virus identification quality to obtain an optimized sensor array.

[0053] This invention discloses a method and system for identifying viral diseases in cucurbit vegetables based on a sensor array. The method generates a dynamic heat map by collecting insect activity data from the target planting area, and constructs a viral disease transmission distribution map based on an insect vector transmission probability model, thereby deploying an initial sensor array. It then acquires real-time plant physiological parameters and environmental data through multiple sensors to construct a viral disease identification model that integrates insect vector activity characteristics, enabling early disease diagnosis. Based on the identification results, the sensor array performance is evaluated, and an adaptive optimization algorithm is used to dynamically adjust the array layout, forming an optimized network that balances monitoring accuracy and resource efficiency. This significantly improves the sensitivity and accuracy of viral disease detection, reduces monitoring costs, and provides a digital solution for the prevention and control of diseases in cucurbit crops. Attached Figure Description

[0054] Figure 1A flowchart of a method for identifying viral diseases in cucurbit vegetables based on a sensor array, according to the present invention, is shown.

[0055] Figure 2 The flowchart illustrating the invention for constructing a heat map of insect activity in a target planting area is shown.

[0056] Figure 3 The flowchart illustrating the identification results of cucurbit virus diseases obtained by the present invention is shown.

[0057] Figure 4 A block diagram of a cucurbit vegetable virus disease identification system based on a sensor array is shown. Detailed Implementation

[0058] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0059] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0060] Figure 1 The flowchart of a method for identifying viral diseases in cucurbit vegetables based on a sensor array, according to the present invention, is shown.

[0061] like Figure 1 As shown, the first aspect of the present invention provides a method for identifying viral diseases in cucurbit vegetables based on a sensor array, comprising:

[0062] S102, Obtain insect activity data in the target planting area of ​​cucurbit vegetables, and construct an insect activity heat map of the target planting area based on the insect activity data;

[0063] S104, construct a vector-borne transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area based on the insect activity heat map, and construct an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the vector-borne transmission probability distribution map;

[0064] S106, acquire virus disease monitoring data from each sensor, construct a virus disease identification model for cucurbit vegetables, identify the virus disease monitoring data according to the virus disease identification model, and obtain the virus disease identification result for cucurbit vegetables;

[0065] S108, determine the virus identification quality of the initial sensor array based on the identification result, and adjust the initial sensor array based on the virus identification quality to obtain an optimized sensor array.

[0066] It should be noted that the construction of insect activity heatmaps can capture the distribution density and migration patterns of insect vectors, accurately locate potential source areas and diffusion directions of viruses, and overcome the spatial and temporal limitations of traditional manual inspections. Secondly, by deploying the initial sensor array in conjunction with the insect vector transmission probability distribution map, hyperspectral imaging sensors are prioritized in areas with high transmission probability, effectively enhancing the monitoring sensitivity of high-risk areas and avoiding resource waste from uniform deployment and insufficient monitoring in key areas. Subsequently, through the integration and analysis of multi-source sensor data using a virus disease identification model, plant spectral reflectance characteristics and insect vector activity parameters are simultaneously analyzed, enabling refined identification of virus disease types, infection stages, and severity, significantly improving classification accuracy. Finally, based on the spatial quality assessment of the identification results, an intelligent optimization algorithm is used to dynamically adjust the sensor layout: increasing sensor density and optimizing monitoring parameters in areas with high transmission risk but low identification quality, and reducing redundant equipment in low-risk, high-precision areas, thereby constructing an adaptive monitoring network that optimizes resource allocation efficiency while ensuring full coverage of the core virus transmission area.

[0067] Figure 2 The flowchart illustrating the construction of a heat map of insect activity in a target planting area according to the present invention is shown.

[0068] According to an embodiment of the present invention, the step of obtaining insect activity data in the target planting area of ​​cucurbit vegetables and constructing an insect activity heat map of the target planting area based on the insect activity data specifically includes:

[0069] S202, acquire historical insect infestation record data of the target planting area of ​​cucurbit vegetables, determine the active insect types in the target planting area based on the historical insect infestation record data, acquire standard image data of the active insect types, and label the standard image data with insect names to obtain labeled image data;

[0070] S204, Construct an insect recognition model based on a convolutional neural network, construct convolutional layers, pooling layers and fully connected layers of the insect recognition model, and construct a cross-entropy loss function and an Adam optimizer, and import the labeled image data into the insect recognition model for training;

[0071] S206, the target planting area is divided into N sub-regions of preset size, and image data of each sub-region is acquired based on a high-definition camera device during a preset time period. The image data is then imported into the insect recognition model for insect recognition to obtain insect activity data for each sub-region. The insect activity data includes insect type and insect quantity.

[0072] S208: Based on the kernel density estimation algorithm, each sub-region is regarded as an observation point, the number of insects is used as the weight value of the observation point, a Gaussian kernel function is selected and the bandwidth parameter of the Gaussian kernel function is determined, and a continuous spatial insect density distribution map is constructed in the target planting area according to the kernel density estimation algorithm.

[0073] S210, Based on the insect density distribution map, the insect density at each location is color-mapped to construct a heat map of insect activity in the target planting area over a preset time period.

[0074] It should be noted that an insect type database was established based on historical insect infestation data and standard image annotations. By combining the convolutional-pooling-fully connected layer structure of a convolutional neural network (CNN) model with optimized cross-entropy loss function, high-precision identification of insect species under complex lighting and shading conditions in the field was achieved, solving the problems of high misjudgment rate and poor timeliness in traditional manual counting. Secondly, by dynamically acquiring image data in different regions using high-definition cameras and inputting it into the trained insect identification model, accurate distribution data of specific insect types and quantities in each sub-region can be obtained in real time, overcoming the spatial limitations of traditional sampling surveys. Furthermore, a kernel density estimation algorithm was used to convert discrete sub-region observation data into a continuous spatial insect density distribution map. Through dynamic optimization of the Gaussian kernel function bandwidth parameter, the changes in insect vector density gradients and migration and aggregation characteristics were accurately characterized, overcoming the shortcomings of traditional grid statistical methods in expressing spatial continuity. Finally, an intuitive insect activity heatmap was generated through color mapping, dynamically reflecting the peak period of insect vector activity and core diffusion paths, providing high spatiotemporal resolution dynamic data support for modeling the probability of viral disease transmission, forming a fully automated monitoring system from data collection to visualization analysis.

[0075] According to an embodiment of the present invention, the step of constructing a vector-borne transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area based on the insect activity heat map, and constructing an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the vector-borne transmission probability distribution map, specifically includes:

[0076] Data on viral disease incidence in cucurbit vegetables in the target planting area during the preset time period are obtained. The data includes the location of the disease, the type of viral disease, the severity of the disease, and the spread path of the disease.

[0077] Spatial overlay analysis was performed on the insect types, insect densities, and disease severity in the corresponding sub-regions of the insect activity heatmaps. Pearson correlation coefficients between the density of each insect type and the severity of disease were calculated, and a correlation coefficient matrix was constructed. Based on the correlation coefficient matrix, target insect types with a correlation coefficient greater than a preset value for viral disease transmission were selected.

[0078] Based on the insect activity heatmap, the distribution density data of the target insect type in the target planting area is extracted. Combined with the directional characteristics of the spread path of the disease location, a virus transmission and diffusion model based on the migration path of the target insect is constructed.

[0079] The virus transmission probability weight of each sub-region is calculated based on the virus transmission and diffusion model. Sub-regions that are closer to the onset of disease and have a higher density of target insects are given a higher transmission probability weight.

[0080] Spatial interpolation is performed on the propagation probability weights of each sub-region to generate a continuous probability surface covering the target planting area. Based on the probability surface threshold, high, medium and low propagation risk areas are divided to construct an insect vector propagation probability distribution map.

[0081] The density of hyperspectral imaging sensors in each sub-region of the target planting area is determined based on the vector propagation probability distribution map. The placement location is determined based on the density. An initial sensor array for identifying viral diseases in cucurbit vegetables in the target planting area is constructed based on the placement location.

[0082] It should be noted that in the target planting areas of cucurbit vegetables, insect activity can spread viruses to these areas, making insect-borne virus transmission a significant route for viral infection in cucurbit vegetables. Therefore, through spatial overlay analysis of insect activity heatmaps and viral disease incidence data, target insect vector types significantly associated with virus transmission are identified, eliminating interference from non-transmission insects and ensuring the biological rationality of the model construction. Secondly, combining the density distribution of target insect vectors with the characteristics of disease spread paths, a virus transmission and diffusion model is constructed, quantifying the transmission probability weights of different sub-regions and overcoming the limitations of traditional uniform deployment that ignores the spatiotemporal correlation between insect vector migration and disease spread. By generating continuous probability surfaces through spatial interpolation and classifying risk levels, abstract insect vector activity data is transformed into a visualized virus transmission heatmap, accurately identifying the frontier areas of virus spread and potentially high-incidence areas. Based on this, the differentiated deployment of hyperspectral imaging sensors is dynamically guided: sensor nodes are densified in high-probability transmission areas to enhance spectral data acquisition density, while sparse deployment is used in low-risk areas to reduce resource redundancy, forming a monitoring network highly matched to the virus transmission path. This scheme not only enhances the sensor array's ability to capture early infection characteristics of viral diseases, but also provides real-time data support for blocking virus spread by dynamically tracking insect-driven propagation trends. The virus transmission and diffusion model is based on the density distribution data of target insects within the planting area, generating a continuous insect density field through spatial interpolation. Simultaneously, the system analyzes the spatial distribution characteristics of viral disease outbreak points, extracting the dominant direction of disease spread. Based on this, the model comprehensively considers three key factors: insect density, distance attenuation effect at the outbreak point, and the degree of alignment with the propagation direction. Distance attenuation uses a nonlinear attenuation function to characterize the weakening of virus transmission with increasing distance, while the degree of alignment with the direction quantifies the consistency between the propagation path and the main spread direction through angle difference calculations. A multi-factor weighted fusion algorithm integrates these elements into a unified propagation probability calculation system, and a spatial anisotropy correction mechanism is introduced to reflect the differences in propagation in different directions. The final generated model can dynamically simulate the spread of viruses through insect vectors in three-dimensional space. The insect-borne transmission probability distribution map is a visualized heatmap reflecting the risk level of viral disease transmission at different locations within the planting area.

[0083] Figure 3 The flowchart illustrating the identification results of cucurbit virus diseases obtained by the present invention is shown.

[0084] According to an embodiment of the present invention, the step of acquiring viral disease monitoring data from each sensor, constructing a viral disease identification model for cucurbit vegetables, and identifying the viral disease monitoring data based on the viral disease identification model to obtain the cucurbit viral disease identification result specifically involves:

[0085] S302, obtain the appearance morphology data of plants with different types of viral diseases at different growth stages, obtain the spectral reflectance data of the appearance morphology, extract the spectral features of the spectral reflectance data based on the PCA algorithm, integrate the spectral features of the same type of viral disease, and construct the spectral feature vector of plants with viral diseases.

[0086] S304, Label the spectral feature vector with the type and severity of the viral disease to obtain training set label spectral feature data;

[0087] S306, a virus disease classification model is constructed based on the support vector machine algorithm. A Gaussian kernel function is selected and the hierarchical loss function and L2 regularization term are defined. The sequence minimum optimization algorithm is used to train the model on the label spectral feature data of the training set. The kernel function parameters and regularization coefficients are adjusted through cross-validation until the classification accuracy exceeds the preset threshold.

[0088] S308, the trained virus disease classification model is deployed to each sensor node in each of the initial sensor arrays, the spectral reflectance data monitored by the sensors is acquired in real time and the spectral features are extracted, the extracted spectral features are input into the virus disease classification model to predict the virus disease type and severity level, the prediction results of all sensors are summarized and mapped to the corresponding sub-regions, and the virus disease identification results of the target planting area are generated.

[0089] It should be noted that by constructing a virus disease classification model based on the support vector machine algorithm, and using kernel functions to map high-dimensional spectral data to the feature space, the problem of spectral nonlinear separability, which is difficult to handle by traditional methods, is effectively solved, enabling accurate differentiation of different viral disease types and their severity levels. By employing a multi-classification strategy and ordered regression method, the system achieves accurate identification of multiple viral diseases and their different disease stages, and is particularly adept at handling the challenge of identifying virus types with similar spectral features. The model's built-in regularization mechanism and soft-margin classification design significantly improve its anti-interference ability, enabling it to adapt to data collection conditions under complex field environments. Simultaneously, the model supports online learning and incremental updates, continuously optimizing classification performance as new data accumulates. The morphological data includes plant dwarfing, leaf yellowing, mosaic, curling, deformity, delayed plant development, and fruit deformity; the spectral features include absorption peaks, reflection peaks, emission peaks, wavelength position, intensity, width, shape, continuity, slope, and characteristic spectral bands.

[0090] According to an embodiment of the present invention, the step of determining the virus identification quality of the initial sensor array based on the identification result, and adjusting the initial sensor array based on the virus identification quality to obtain an optimized sensor array, specifically involves:

[0091] The actual viral disease infection data of cucurbit vegetables in each sub-region of the target planting area is obtained. The actual viral disease infection data is compared with the viral disease identification results to determine the accuracy and false negative rate of the initial sensor array for viral disease identification in each sub-region.

[0092] The identification quality of cucurbit vegetables in each sub-region was evaluated based on the accuracy and false negative rate of the virus disease identification, and the initial sensor array's virus disease identification quality data for each sub-region was obtained.

[0093] The virus disease identification quality data is clustered based on the BI RCH clustering algorithm, dividing the target planting area into high identification quality area, medium identification quality area and low identification quality area, and generating a spatially distributed identification quality cluster map.

[0094] The identification quality clustering map and the vector transmission probability distribution map are spatially overlaid to extract the proportion of the area covered by the high identification quality area in the high-risk area of ​​vector transmission, and the identification coverage rate of the vector transmission risk area is calculated.

[0095] It should be noted that the initial sensor array may not be able to fully monitor and identify all areas at risk of vector-borne disease transmission due to installation angle or location issues. By using the BI RCH clustering algorithm to cluster viral disease identification quality data and constructing an identification quality clustering map, rapid hierarchical clustering of large-scale farmland monitoring data can be achieved. By dynamically adjusting the threshold parameters of the clustering feature tree, the boundaries of high, medium, and low identification quality areas can be accurately defined. The generated identification quality clustering map can intuitively reflect the weak and strong areas of the sensor array's monitoring effectiveness. After spatial overlay analysis combined with the vector-borne disease transmission probability distribution map, the effective coverage of the sensor monitoring network in high-risk areas can be quantitatively evaluated by calculating the identification coverage rate of vector-borne disease transmission risk areas.

[0096] If the coverage rate of the vector-borne transmission risk area identification is less than a preset threshold, then sensor deployment defect analysis is performed on the uncovered high-risk vector-borne transmission sub-regions to obtain the mean value of virus transmission probability weight and the corresponding virus disease identification quality score of the defective sub-regions.

[0097] When the average virus propagation probability weight of the defective sub-region is greater than the preset risk threshold and the virus disease identification quality score is lower than the preset quality threshold, it is determined to be a sensor deployment failure area. The ant colony optimization algorithm is used to re-plan the sensor deployment path in the failure area, and the density of hyperspectral imaging sensors is increased at the peak point of the virus propagation probability weight to obtain the first optimization strategy.

[0098] When the average weight of the virus transmission probability in the defective sub-region is less than the preset risk threshold but the identification quality score is lower than the preset quality threshold, it is determined to be a sensor monitoring blind zone. The monitoring angle and spectral acquisition frequency of the sensor in the blind zone are adjusted according to the migration direction of the target insect in the insect activity heat map to obtain the second optimization strategy.

[0099] The initial sensor array is adjusted according to the first optimization strategy and the second optimization strategy to obtain an optimized sensor array.

[0100] It should be noted that when the average virus propagation probability weight of a defective sub-region is greater than a preset risk threshold and the identification quality score is lower than a preset quality threshold, it indicates that the region is located in the core path of insect-driven virus spread. However, due to insufficient sensor deployment density, misalignment of node distribution with insect activity hotspots, or mismatch in equipment performance parameters, high-risk areas are not effectively monitored, forming a critical vulnerability in the monitoring network. Therefore, an intelligent optimization mechanism simulating insect foraging paths is used through an ant colony optimization algorithm to generate sensor deployment paths that highly match the virus propagation probability weight distribution in areas where sensor deployment fails. Hyperspectral imaging sensor nodes are prioritized for densification at the weight peak points at the forefront of virus spread. This strategy breaks through the rigid constraints of traditional grid-based deployment, achieving precise tilting of sensor resources towards the core virus propagation channel, effectively improving the monitoring sensitivity and data integrity of high-risk areas, and blocking the risk of missed virus spread due to monitoring blind spots. When the average virus transmission probability weight of a defective sub-region is lower than the risk threshold but the identification quality is still substandard, it indicates that although the region is not the current main path of virus spread, data quality deterioration is caused by sensor monitoring angle deviation, mismatch between spectral acquisition frequency and insect vector activity rhythm, or environmental interference (such as plant shading), forming a non-risk-oriented monitoring blind spot. For monitoring blind spots outside the main risk path, the system dynamically adjusts the sensor monitoring angle based on the insect vector migration direction revealed by the insect activity heatmap (e.g., aligning the spectral acquisition angle with the insect migration direction) and simultaneously increases the spectral acquisition frequency to match the peak activity period of the target insects. This strategy, through the spatiotemporal coupling optimization of sensor parameters and insect vector behavioral characteristics, significantly enhances the ability to capture sporadic transmission events and low-density insect vector activity, reduces data loss caused by equipment response lag or monitoring direction deviation, and achieves a balanced improvement in monitoring efficiency across the entire region.

[0101] According to an embodiment of the present invention, the clustering operation on the virus disease identification quality data based on the BIRCH clustering algorithm specifically includes:

[0102] The viral disease identification quality data is subjected to data standardization processing, and the feature values ​​of each dimension are transformed into zero mean and unit variance distributions to obtain standardized viral disease data.

[0103] The BIRCH clustering algorithm is introduced to calculate the branching factor threshold based on the data dimension and sample number of the standardized viral disease data, and to determine the maximum branching factor and sub-cluster diameter threshold of the clustering feature tree.

[0104] The clustering feature tree structure is initialized based on the maximum branching factor and the sub-cluster diameter threshold, and a root node containing an empty clustering feature vector is created. The clustering feature vector consists of the number of samples, a linear summation vector, and a sum of squares scalar.

[0105] Traverse each data point in the standardized viral disease data, starting from the root node and searching downwards along the tree structure. Calculate the Euclidean distance between the current data point and the cluster feature vectors of each sub-cluster, and select the sub-cluster path with the closest distance for recursive access.

[0106] If the radius of a subcluster in a leaf node does not exceed the subcluster diameter threshold, the current data point is merged into the subcluster and the clustering feature vector is updated; otherwise, a new subcluster is created and the branching factor limit is checked. If the branching factor limit is exceeded, the node is split to generate a new leaf node.

[0107] After all data points have been inserted, the clustering feature tree is condensed, and the subclusters in adjacent leaf nodes that meet the diameter threshold are merged to form coarse-grained clusters.

[0108] Based on the coarse-grained clustering results, the clustering feature vectors of all sub-clusters are extracted as a new dataset. A hierarchical clustering algorithm is used to perform global clustering operations. The distance matrix between sub-clusters is calculated and the final number of clusters is determined by cutting the dendrogram.

[0109] The viral disease identification quality data is mapped to the global clustering results according to the sub-cluster affiliation relationship, thus obtaining the clustering results of the viral disease identification quality data.

[0110] It should be noted that after eliminating the dimensional differences of multi-dimensional quality indicators through data standardization, the BIRCH clustering algorithm is used to construct a clustering feature tree by calculating the branching factor threshold and the sub-cluster diameter threshold. This adaptively divides the data into dense and sparse regions, effectively solving the overfitting or underfitting problems of traditional algorithms for non-uniformly distributed data. Incremental data clustering is achieved by combining recursive insertion and node splitting mechanisms, significantly reducing the computational cost in dynamic monitoring scenarios. Simultaneously, redundant sub-clusters are compressed through cluster reduction operations, and the tree diagram cutting strategy of hierarchical clustering accurately identifies spatial heterogeneity features, forming highly discriminative quality partitions. The BIRCH clustering algorithm achieves efficient processing of large-scale viral disease identification quality data by constructing a dynamically branched clustering feature tree. This algorithm automatically controls the clustering granularity using the sub-cluster diameter threshold, effectively identifying data distribution characteristics and significantly reducing computational complexity while maintaining accuracy.

[0111] Figure 4A block diagram of a cucurbit vegetable virus disease identification system based on a sensor array is shown.

[0112] A second aspect of the present invention also provides a cucurbit vegetable virus disease identification system 4 based on a sensor array. The system includes a memory 41 and a processor 42. The memory includes a program for a cucurbit vegetable virus disease identification method based on a sensor array. When the processor executes the program, the program performs the following steps:

[0113] Data on insect activity in the target planting area of ​​cucurbit vegetables is obtained, and a heat map of insect activity in the target planting area is constructed based on the insect activity data;

[0114] Based on the insect activity heat map, construct an insect vector transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area, and construct an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the insect vector transmission probability distribution map;

[0115] The virus disease monitoring data of each sensor is acquired, a virus disease identification model for cucurbit vegetables is constructed, and the virus disease monitoring data is identified according to the virus disease identification model to obtain the virus disease identification results for cucurbit vegetables;

[0116] The virus identification quality of the initial sensor array is determined based on the identification results, and the initial sensor array is adjusted based on the virus identification quality to obtain an optimized sensor array.

[0117] This invention discloses a method and system for identifying viral diseases in cucurbit vegetables based on a sensor array. The method generates a dynamic heat map by collecting insect activity data from the target planting area, and constructs a viral disease transmission distribution map based on an insect vector transmission probability model, thereby deploying an initial sensor array. It then acquires real-time plant physiological parameters and environmental data through multiple sensors to construct a viral disease identification model that integrates insect vector activity characteristics, enabling early disease diagnosis. Based on the identification results, the sensor array performance is evaluated, and an adaptive optimization algorithm is used to dynamically adjust the array layout, forming an optimized network that balances monitoring accuracy and resource efficiency. This significantly improves the sensitivity and accuracy of viral disease detection, reduces monitoring costs, and provides a digital solution for the prevention and control of diseases in cucurbit crops.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0119] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0120] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0121] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0123] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying a viral disease of a melon vegetable based on a sensor array construction, characterized by, Includes the following steps: Data on insect activity in the target planting area of ​​cucurbit vegetables is obtained, and a heat map of insect activity in the target planting area is constructed based on the insect activity data; Based on the insect activity heat map, construct an insect vector transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area, and construct an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the insect vector transmission probability distribution map; The virus disease monitoring data of each sensor is acquired, a virus disease identification model for cucurbit vegetables is constructed, and the virus disease monitoring data is identified according to the virus disease identification model to obtain the virus disease identification results for cucurbit vegetables; Based on the identification results, the virus identification quality of the initial sensor array is determined. The initial sensor array is then adjusted based on the virus identification quality to obtain an optimized sensor array, specifically as follows: The actual viral disease infection data of cucurbit vegetables in each sub-region of the target planting area is obtained. The actual viral disease infection data is compared with the viral disease identification results to determine the accuracy and false negative rate of the initial sensor array for viral disease identification in each sub-region. The identification quality of cucurbit vegetables in each sub-region was evaluated based on the accuracy and false negative rate of the virus disease identification, and the initial sensor array's virus disease identification quality data for each sub-region was obtained. The virus disease identification quality data is clustered based on the BIRCH clustering algorithm, dividing the target planting area into high identification quality area, medium identification quality area and low identification quality area, and generating a spatially distributed identification quality cluster map. The identification quality clustering map and the vector transmission probability distribution map are spatially overlaid to extract the proportion of the area covered by the high identification quality area in the high-risk area of ​​vector transmission, and the identification coverage rate of the vector transmission risk area is calculated. If the coverage rate of the vector-borne transmission risk area identification is less than a preset threshold, then sensor deployment defect analysis is performed on the uncovered high-risk vector-borne transmission sub-regions to obtain the mean value of virus transmission probability weight and the corresponding virus disease identification quality score of the defective sub-regions. When the average virus propagation probability weight of the defective sub-region is greater than the preset risk threshold and the virus disease identification quality score is lower than the preset quality threshold, it is determined to be a sensor deployment failure area. The ant colony optimization algorithm is used to re-plan the sensor deployment path in the failure area, and the density of hyperspectral imaging sensors is increased at the peak point of the virus propagation probability weight to obtain the first optimization strategy. When the average weight of the virus transmission probability in the defective sub-region is less than the preset risk threshold but the identification quality score is lower than the preset quality threshold, it is determined to be a sensor monitoring blind zone. The monitoring angle and spectral acquisition frequency of the sensor in the blind zone are adjusted according to the migration direction of the target insect in the insect activity heat map to obtain the second optimization strategy. The initial sensor array is adjusted according to the first optimization strategy and the second optimization strategy to obtain an optimized sensor array.

2. The method according to claim 1, wherein the method is characterized by, The process of acquiring insect activity data in the target planting area of ​​cucurbit vegetables and constructing an insect activity heatmap of the target planting area based on the insect activity data specifically involves: Historical insect infestation records of the target planting area for cucurbit vegetables are obtained. Based on the historical insect infestation records, the types of active insects in the target planting area are determined. Standard image data of the active insect types are obtained. Insect names are labeled on the standard image data to obtain labeled image data. An insect recognition model is constructed based on a convolutional neural network. The model consists of convolutional layers, pooling layers, and fully connected layers. A cross-entropy loss function and an Adam optimizer are also constructed. The labeled image data is then imported into the insect recognition model for training. The target planting area is divided into N sub-regions of a preset size. Image data of each sub-region is acquired within a preset time period using a high-definition camera. The image data is then imported into the insect recognition model for insect recognition to obtain insect activity data for each sub-region. The insect activity data includes insect type and insect quantity. Based on the kernel density estimation algorithm, each sub-region is regarded as an observation point, the number of insects is used as the weight value of the observation point, a Gaussian kernel function is selected and the bandwidth parameter of the Gaussian kernel function is determined, and a continuous spatial insect density distribution map is constructed in the target planting area according to the kernel density estimation algorithm. Based on the insect density distribution map, the insect density at each location is color-mapped to construct a heat map of insect activity in the target planting area over a preset time period.

3. The method according to claim 2, wherein the method is characterized by, The step of constructing a vector-borne transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area based on the insect activity heat map, and constructing an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the vector-borne transmission probability distribution map, specifically involves: Data on viral disease incidence in cucurbit vegetables in the target planting area during the preset time period are obtained. The data includes the location of the disease, the type of viral disease, the severity of the disease, and the spread path of the disease. Spatial overlay analysis was performed on the insect types, insect densities, and disease severity in the corresponding sub-regions of the insect activity heatmaps. Pearson correlation coefficients between the density of each insect type and the severity of disease were calculated, and a correlation coefficient matrix was constructed. Based on the correlation coefficient matrix, target insect types with a correlation coefficient greater than a preset value for viral disease transmission were selected. Based on the insect activity heatmap, the distribution density data of the target insect type in the target planting area is extracted. Combined with the directional characteristics of the spread path of the disease location, a virus transmission and diffusion model based on the migration path of the target insect is constructed. The virus transmission probability weight of each sub-region is calculated based on the virus transmission and diffusion model. Sub-regions that are closer to the onset of disease and have a higher density of target insects are given a higher transmission probability weight. Spatial interpolation is performed on the propagation probability weights of each sub-region to generate a continuous probability surface covering the target planting area. Based on the probability surface threshold, high, medium and low propagation risk areas are divided to construct an insect vector propagation probability distribution map. The density of hyperspectral imaging sensors in each sub-region of the target planting area is determined based on the vector propagation probability distribution map. The placement location is determined based on the density. An initial sensor array for identifying viral diseases in cucurbit vegetables in the target planting area is constructed based on the placement location.

4. The method according to claim 1, wherein the method is characterized by, The process involves acquiring viral disease monitoring data from each sensor, constructing a viral disease identification model for cucurbit vegetables, and identifying the viral disease monitoring data based on the viral disease identification model to obtain the viral disease identification results for cucurbit vegetables. Specifically: The appearance morphology data of plants with different types of viral diseases at different growth stages are obtained, and the spectral reflectance data of the appearance morphology are obtained. The spectral features of the spectral reflectance data are extracted based on the PCA algorithm. The spectral features of the same type of viral disease are integrated to construct the spectral feature vector of plants with viral diseases. The spectral feature vectors are labeled with the viral disease type and severity to obtain training set labeled spectral feature data; A virus disease classification model is constructed based on the support vector machine algorithm. A Gaussian kernel function is selected and a hinge loss function and an L2 regularization term are defined. The sequence minimum optimization algorithm is used to train the model on the label spectral feature data of the training set. The kernel function parameters and regularization coefficients are adjusted through cross-validation until the classification accuracy exceeds a preset threshold. The trained viral disease classification model is deployed to each sensor node in the initial sensor array. The spectral reflectance data monitored by the sensors is acquired in real time and spectral features are extracted. The extracted spectral features are input into the viral disease classification model to predict the type and severity level of the viral disease. The prediction results of all sensors are summarized and mapped to the corresponding sub-regions to generate the viral disease identification results of the target planting area.

5. The method for identifying viral diseases of cucurbit vegetables based on a sensor array as described in claim 1, characterized in that, The clustering operation on the virus disease identification quality data based on the BIRCH clustering algorithm is specifically as follows: The viral disease identification quality data is subjected to data standardization processing, and the feature values ​​of each dimension are transformed into zero mean and unit variance distributions to obtain standardized viral disease data. The BIRCH clustering algorithm is introduced to calculate the branching factor threshold based on the data dimension and sample number of the standardized viral disease data, and to determine the maximum branching factor and sub-cluster diameter threshold of the clustering feature tree. The clustering feature tree structure is initialized based on the maximum branching factor and the sub-cluster diameter threshold, and a root node containing an empty clustering feature vector is created. The clustering feature vector consists of the number of samples, a linear summation vector, and a sum of squares scalar. Traverse each data point in the standardized viral disease data, starting from the root node and searching downwards along the tree structure. Calculate the Euclidean distance between the current data point and the cluster feature vectors of each sub-cluster, and select the sub-cluster path with the closest distance for recursive access. If the radius of a subcluster in a leaf node does not exceed the subcluster diameter threshold, the current data point is merged into the subcluster and the clustering feature vector is updated; otherwise, a new subcluster is created and the branching factor limit is checked. If the branching factor limit is exceeded, the node is split to generate a new leaf node. After all data points have been inserted, the clustering feature tree is condensed, and the subclusters in adjacent leaf nodes that meet the diameter threshold are merged to form coarse-grained clusters. Based on the coarse-grained clustering results, the clustering feature vectors of all sub-clusters are extracted as a new dataset. A hierarchical clustering algorithm is used to perform global clustering operations. The distance matrix between sub-clusters is calculated and the final number of clusters is determined by cutting the dendrogram. The viral disease identification quality data is mapped to the global clustering results according to the sub-cluster affiliation relationship, thus obtaining the clustering results of the viral disease identification quality data.

6. A virus disease identification system for cucurbit vegetables based on a sensor array, characterized in that, The cucurbit vegetable virus disease identification system based on a sensor array includes a storage device and a processor. The storage device includes a cucurbit vegetable virus disease identification method program based on a sensor array. When the processor executes the cucurbit vegetable virus disease identification method program based on a sensor array, it performs the following steps: Data on insect activity in the target planting area of ​​cucurbit vegetables is obtained, and a heat map of insect activity in the target planting area is constructed based on the insect activity data; Based on the insect activity heat map, construct an insect vector transmission probability distribution map of cucurbit vegetable viral diseases in the target planting area, and construct an initial sensor array for identifying cucurbit vegetable viral diseases in the target planting area based on the insect vector transmission probability distribution map; The virus disease monitoring data of each sensor is acquired, a virus disease identification model for cucurbit vegetables is constructed, and the virus disease monitoring data is identified according to the virus disease identification model to obtain the virus disease identification results for cucurbit vegetables; Based on the identification results, the virus identification quality of the initial sensor array is determined. The initial sensor array is then adjusted based on the virus identification quality to obtain an optimized sensor array, specifically as follows: The actual viral disease infection data of cucurbit vegetables in each sub-region of the target planting area is obtained. The actual viral disease infection data is compared with the viral disease identification results to determine the accuracy and false negative rate of the initial sensor array for viral disease identification in each sub-region. The identification quality of cucurbit vegetables in each sub-region was evaluated based on the accuracy and false negative rate of the virus disease identification, and the initial sensor array's virus disease identification quality data for each sub-region was obtained. The virus disease identification quality data is clustered based on the BIRCH clustering algorithm, dividing the target planting area into high identification quality area, medium identification quality area and low identification quality area, and generating a spatially distributed identification quality cluster map. The identification quality clustering map and the vector transmission probability distribution map are spatially overlaid to extract the proportion of the area covered by the high identification quality area in the high-risk area of ​​vector transmission, and the identification coverage rate of the vector transmission risk area is calculated. If the coverage rate of the vector-borne transmission risk area identification is less than a preset threshold, then sensor deployment defect analysis is performed on the uncovered high-risk vector-borne transmission sub-regions to obtain the mean value of virus transmission probability weight and the corresponding virus disease identification quality score of the defective sub-regions. When the average virus propagation probability weight of the defective sub-region is greater than the preset risk threshold and the virus disease identification quality score is lower than the preset quality threshold, it is determined to be a sensor deployment failure area. The ant colony optimization algorithm is used to re-plan the sensor deployment path in the failure area, and the density of hyperspectral imaging sensors is increased at the peak point of the virus propagation probability weight to obtain the first optimization strategy. When the mean value of the virus transmission probability weight of the defective sub-region is less than the preset risk threshold but the recognition quality score is lower than the preset quality threshold, it is determined as the sensor monitoring blind area, and the monitoring angle and spectral acquisition frequency of the sensors in the blind area are adjusted according to the migration direction of the target insects in the insect activity heat map to obtain the second optimization strategy; Adjust the initial sensor array according to the first optimization strategy and the second optimization strategy to obtain an optimized sensor array.

Images