Intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning

Abstract

The present invention relates to the technical field of crop disease and insect pest monitoring, and in particular provides an intelligent crop disease and insect pest monitoring system based on image recognition and machine learning. The system comprises: a benchmark model generation subsystem for acquiring a continuous state stream within the crop growth cycle through a multispectral imaging array; a micro-field comparison of the real-time image of the continuous state stream with an ideal morphological benchmark vector, outputting a biological stress difference map; a mutually exclusive association engine subsystem for purifying the pathological response map, inputting it into a pathological feature dissociation network, and outputting a set of environmental immune disease markers; and a pathogen source trajectory deduction subsystem for analyzing the initial infection focus coordinates, combined with a real-time wind direction vector field and a plant density topology map, through a biological diffusion dynamic chain. The present invention solves the problem of high false positive rates in complex farmland scenarios through a dual denoising mechanism of growth rhythm decoupling and environmental interference stripping, ultimately achieving a leap in monitoring dimensions from single-point identification to group prevention and control.

Description

Technical Field

[0001] The present invention relates to the technical field of crop disease and insect pest monitoring, and in particular to an intelligent crop disease and insect pest monitoring system based on image recognition and machine learning. Background Art

[0002] With the rapid development of global agriculture, the monitoring and control of crop pests and diseases has become a critical component in ensuring food security and sustainable agricultural development. Traditional pest and disease monitoring methods rely primarily on manual field surveys, which are not only time-consuming and labor-intensive but also subject to subjective judgment, making them prone to misjudgments and missed detections. Furthermore, the overuse of chemical pesticides has led to environmental pollution and ecological damage. In recent years, with advances in computer vision and machine learning, crop pest and disease monitoring technologies based on image recognition and intelligent algorithms have become a research hotspot. These technologies collect images of crop leaves, stems, and other parts of the body and use algorithms to automatically identify the type and severity of pests and diseases, providing a scientific basis for precision agriculture. However, existing technologies still have many shortcomings in terms of real-time performance, accuracy, and adaptability, necessitating a more efficient and reliable intelligent monitoring solution.

[0003] Prior art 1, Chinese patent, publication number CN118134012A discloses a crop disease and pest control program recommendation system based on deep learning, including a crop monitoring module, an environmental perception module, a disease and pest control database, a disease and pest prediction module, a program generation module, a decision visualization module, an adaptive learning module, a community sharing platform, and a remote collaboration module. Although it can find the global optimal crop disease and pest control strategy, improve the overall effect of the program, help the system better adapt to geographical and seasonal changes, cope with the uncertainty in agricultural production, improve the robustness of the control program, realize dynamic optimization of the strategy, reduce data search time, increase data access speed, reduce frequent access to the database, and improve system stability; however, the disease and pest control program recommendation system is provided, but it relies on a static database and prediction model, and does not establish a dynamic mapping relationship between crop growth benchmarks and pathological characteristics, resulting in insufficient early disease and pest detection capabilities and an inability to distinguish between environmental stress and real pathological signals.

[0004] Prior art two, Chinese patent publication number CN119000673A, discloses a remote sensing monitoring device for crop pests and diseases and a monitoring method thereof. The device comprises a main frame, a control chamber mounted on the top outer wall of the main frame, a monitor mounted on the top of the control chamber, a signal receiver mounted on the middle outer wall of the main frame, and a data processor mounted on the middle outer wall of the main frame. Pest and disease treatment mechanisms are symmetrically arranged on the left and right sides of the main frame, and a monitoring and protection mechanism is provided on the lower surface of the monitor. This device directs and modifies the flow of natural wind, forming a spiral flow around the tree trunk. This spiral flow penetrates cracks and rough surfaces in the tree trunk, dislodging hidden pests. While this device can continuously treat pests in the tree trunk, preventing them from inhabiting and reproducing there for extended periods, significantly reducing the risk of pest spread and aggravation, it focuses on the physical removal of pests (e.g., windblowing), lacks the ability to identify diseases, and does not incorporate spatial predictions based on pathogen transmission patterns.

[0005] Prior art three, Chinese patent publication number CN119165833A, discloses a control method for automatic monitoring equipment for crop pests and diseases. Based on field crop data, it mines monitoring links within the growth cycle and, in combination with a set monitoring mode, conducts self-learning on the front-end sensor array of the automatic monitoring equipment. Inspections are conducted based on a preset path, triggering the first sensor array for range monitoring control and the second sensor array for directional monitoring control. While this method addresses the technical issues of insufficient intelligence and flexibility in the control of automatic monitoring equipment in the prior art, and its inability to adapt to different growth periods and diverse pest and disease characteristics in complex environments, resulting in limited accuracy and reliability in pest and disease monitoring, it achieves efficient automated monitoring of pests and diseases and ensures the accuracy and reliability of monitoring data in complex environments. However, while optimizing the sensor control logic, it fails to address the coupling problem of multi-source data (environmental and biological signals), resulting in a high false positive rate.

[0006] Currently, existing technologies 1, 2, and 3 suffer from problems such as delayed early detection, environmental noise interference, missing propagation predictions, and insufficient multimodal data coupling. Therefore, the present invention provides an intelligent crop pest and disease monitoring system based on image recognition and machine learning. Summary of the Invention

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] In one aspect of the present invention, an intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning is provided, comprising:

[0009] The pathogen trajectory deduction subsystem is configured to input the disease marker set into the pathogen transmission dynamics model constructed by the aerosol deposition / root contact probability network of the pathogen in the crop population, and calculate the initial infection focus coordinates; based on the initial infection focus coordinates, combined with the real-time wind direction vector field and plant density topology map, through the biological diffusion dynamic chain.

[0010] In an optional embodiment, the pathogen trajectory deduction subsystem includes:

[0011] The infection source probability field construction component is configured to input the symptom marker set into the pathogen transmission dynamics model; construct an airborne transmission trajectory mechanism based on the airborne transmission characteristics of pathogen spores, and construct an underground contact diffusion mechanism based on the expansion characteristics of soil pathogen hyphae; the airborne transmission trajectory mechanism and the underground contact diffusion mechanism are convolved and superimposed by the spatial symptom density and transmission probability to calculate the coordinates of the initial infection focus;

[0012] The three-dimensional biological diffusion simulation component is configured to input the biological diffusion dynamic chain with the focal coordinates, superimpose the real-time wind direction vector field through the turbulence corrector to generate the air transmission probability flow; superimpose the plant density topology map through the underground resistance estimator to generate the soil contact probability flow; the air transmission probability flow and the soil contact probability flow are output through the cross-medium coupler to output the three-dimensional transmission probability cloud;

[0013] The risk thermal field conversion component is configured as a risk energy level converter that maps the energy of the crop resistance threshold and the pathogen outbreak critical point into a three-dimensional transmission probability cloud input; it marks the over-threshold area in the transmission probability cloud through tolerance critical detection; and generates a disease thermal threat field by weighting the over-threshold area through time domain attenuation.

[0014] In an optional embodiment, the infection source probability field construction component includes:

[0015] The airborne transmission trajectory mechanism construction subcomponent is configured to input the disease marker set into an aerosol deposition network that includes the coupled adsorption effect of pathogen spore particle size and canopy turbulent kinetic energy; generate dynamic deposition trajectories based on the spore air suspension persistence; and quantify the deposition potential energy of spores at each plant site using vortex attachment nuclei.

[0016] The subcomponent constructs the underground contact diffusion mechanism, configured to use a set of disease markers to drive a root contact network that topologically matches the pressure of pathogen hyphae tips with cracks in the root epidermis. The optimal hyphae path set is generated based on the root interlacing density of the overlapping rhizospheres of adjacent plants. The contact potential energy is calculated based on the biochemical reaction rate of hyphae secreting extracellular enzymes to dissolve root waxes.

[0017] The infection focus convolution aggregation sub-component is configured as an infection potential energy aggregator with a dual-mechanism output and input field theory superposition of spatial disease density and propagation potential energy; the sedimentation potential energy and contact potential energy are coupled through the cross-medium potential field to generate a comprehensive infection potential field; the comprehensive infection potential field is located by the extreme value of the three-dimensional potential field gradient change rate, and the initial infection focus coordinates are output.

[0018] In an optional embodiment, the infection focus convolution aggregation subcomponent comprises:

[0019] A potential energy input module is configured as a cross-medium potential field coupler for inputting the sedimentation potential energy and the contact potential energy into the quantitative mechanism of the biochemical chain reaction of the airborne spore sedimentation triggering the secondary infection of the soil;

[0020] The biopotential field fusion module is configured to accelerate the expansion of hyphae through the biochemical pathway of secreting signal molecules through spore germination, converting sedimentation potential energy into a soil infection enhancement coefficient. The contact potential energy activates hyphae invasion through the rhizosphere, inducing crops to release volatile organic compounds that attract spores, and converts it into an airborne transmission enhancement coefficient. The dual enhancement coefficients and the original potential energy are superimposed on the biofield to form a three-dimensional comprehensive infection potential field.

[0021] The potential field gradient extreme value positioning module is configured as a potential energy peak detector based on the principle of critical energy transition of pathogen groups breaking through crop defenses in the three-dimensional comprehensive infection potential field input; the gradient change rate tensor of each point in the potential field space is calculated through the biopotential gradient mechanism of electrochemical potential transition at the junction of healthy tissue and infected tissue; the gradient change rate tensor is locked to the maximum infection potential energy point through the extreme value convergence kernel, and the output three-dimensional coordinates are the coordinates of the initial infection focus.

[0022] In an optional embodiment, the potential field gradient extreme value positioning module includes:

[0023] The infection electrochemical field construction submodule is configured to input the three-dimensional integrated infection potential field into the biopotential gradient operator that responds to the opening and closing of ion channels in the plant cell membrane before and after pathogen invasion. The infection potential field is converted into a tissue potential field through a mechanism that quantifies the transmembrane potential difference between healthy and diseased cells. An ion flux detection core, which follows the diffusion trajectory of pathogen-secreted effector proteins that disrupt cellular ion balance, is used to mark the electrochemical transition interface in the potential field.

[0024] The gradient tensor dynamic generation submodule is configured as a potential differential unit for the spatial conductivity characteristics of ion concentration gradients within the plant vascular system, which is input into the electrochemical transition interface; the plasmodesmata potential conductivity is calculated along the leaf vein vessel direction to determine the anisotropic resistance of cell wall pores to ion flow; and the xylem potential decay rate is calculated along the stem vascular bundle direction to determine the turbulent diffusion effect of pathogenic toxins within the vessel.

[0025] The full-space tensor synthesis submodule is configured as a biofield tensor fusion device that fuses the double differential results through the topological covariance of the electrochemical response of plant tissue to generate a gradient change rate tensor.

[0026] In an optional implementation, the full-space tensor synthesis submodule includes:

[0027] The conductivity orthogonal base construction unit is configured to input the plasmodesmata potential conductivity and the xylem potential attenuation rate into the tissue conductivity orthogonalization unit of the conductivity anisotropy of the spongy tissue and palisade tissue of the plant leaf; the plasmodesmata potential conductivity is calibrated by the leaf vein admittance calibrator based on the spatial modulation effect of the ion flow on the micropore distribution of the mesophyll cell wall, thereby generating a transverse conductivity weight field;

[0028] The three-dimensional covariance computation unit is configured to perform conductance-attenuation coupling at the vascular bundle intersection node, the ion redistribution hub at the petiole-stem junction. It calculates the tensor covariance coefficients using the nodal flow conservation law for charge conservation in local crop tissues under pathogen infection.

[0029] The global tensor generation unit is configured as a biofield tensor synthesis kernel of the tensor covariance coefficient through the covariance derivative rule of the electrochemical parameters of the plant tissue in the Euclidean space, and outputs the gradient change rate tensor.

[0030] In an optional implementation, the three-dimensional covariant computing unit includes:

[0031] The node ion flow balance detection subunit is configured as a node flow conservation detector based on the mass conservation law of bidirectional ion flow at healthy vascular bundle nodes to obtain a theoretical ion flux balance value; and a pathogenic ion interference library based on the leakage rate database of ion channels blocked by specific pathogenic effector proteins is called to generate actual ion leakage.

[0032] The charge conservation constraint solver is configured to input the theoretical ion flux balance value and the actual ion leakage into the conservation constraint solver. The minimum leakage compensation flow is obtained through the ion compensation mechanism from adjacent healthy tissue to the infected node. The minimum leakage compensation flow is converted into a node covariant scaling factor through the infection-compensation balance.

[0033] The global covariance coefficient synthesis subunit is configured as a tensor covariance mapper that inputs the node covariance scale factor into the covariance transmission rule determined by the vascular network topology structure; the spatial weight coefficient of each node is obtained according to the vascular bundle hierarchical transmission law; the spatial weight coefficient and the node covariance scale factor are generated into a tensor covariance coefficient through a hierarchical covariance aggregator.

[0034] In an optional embodiment, the global covariance coefficient synthesizes the ion flow attenuation gradient of the vascular bundle level conduction law of the main stem, branches, and leaf veins of the subunit.

[0035] In an optional embodiment, it also includes: a benchmark model generation subsystem, which is configured to obtain a continuous state stream containing leaf reflection characteristics and branch morphology fluctuation time series data during the crop growth cycle through a multispectral imaging array; the continuous state stream is input into a growth rhythm decoupler that separates environmental interference from biological intrinsic changes, and extracts the ideal morphological benchmark vector of the crop in the current growth stage; the real-time image of the continuous state stream is compared with the ideal morphological benchmark vector for a micro-change field, and a biological stress difference map that only retains the characteristics that deviate from the normal growth rhythm is output.

[0036] In an optional embodiment, it also includes: a mutually exclusive association engine subsystem, which is configured to synchronously input the biological stress difference map with environmental sensor data such as soil ion permeability and canopy microclimate flow; through the environmental interference stripping layer, the noise of environmental factors in the difference map is eliminated to generate a purified pathological response map; the purified pathological response map is input into the pathological feature dissociation network, and an environmental immune disease marker set is output that only contains abnormal features that cannot be explained by environmental factors.

[0037] The benchmark model generation subsystem of the present invention constructs a dynamic benchmark for the ideal growth state of crops through multi-spectral time series data streams, realizing the de-environmental noise removal of growth anomaly detection; the micro-variation field comparison technology upgrades traditional static image recognition to growth rhythm deviation detection, so that the sensitivity of early disease and insect pest response is improved to the subclinical stage. The mutually exclusive association engine subsystem eliminates false alarms caused by non-biological stresses such as meteorology and soil through multi-modal mutually exclusive analysis of environmental sensor data and biological stress characteristics; the environmental immune disease marker set output by the pathological feature dissociation network upgrades traditional disease identification to specific diagnosis at the etiology level. The pathogen trajectory deduction subsystem converts discrete disease markers into spatial epidemiological predictions based on the infection focus positioning of the transmission dynamics model; combining the diffusion dynamic chain of the microclimate field and plant topology, it realizes the deduction from single plant diagnosis to group transmission path, providing spatial coordinate guidance for precise prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0039] Figure 1 This is a block diagram of the intelligent crop disease and insect pest monitoring system based on image recognition and machine learning provided in Example 1 of the present invention;

[0040] Figure 2 A block diagram of a subsystem for generating a benchmark model provided in Example 2 of the present invention;

[0041] Figure 3 This is a block diagram of the mutually exclusive association engine subsystem provided in Example 4 of the present invention;

[0042] Figure 4 This is a block diagram of the pathogen trajectory deduction subsystem provided in Example 5 of the present invention;

[0043] Figure 5 A block diagram of the electronic device provided by the present invention;

[0044] Figure 6 A block diagram of a computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0046] In the following, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature identified with "first," "second," etc., may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0047] In the present invention, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or an integrated one; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. In addition, unless otherwise clearly specified and limited, the term "coupling" should be understood in a broad sense. For example, "coupling" can be a direct electrical connection, such as physical contact and electrical conduction between two components, or it can be understood as the electrical connection between different components in a circuit structure through a physical line that can transmit electrical signals, such as printed circuit board (PCB) copper foil or wire, so as to transmit electrical signals; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in an airless / non-contact manner, such as electrical connection between two components using capacitive coupling to transmit electrical signals.

[0048] In an embodiment of the present invention, directional terms such as "up", "down", "left" and "right" may be defined including but not limited to the orientation relative to the schematic placement of the components in the drawings. It should be understood that these directional terms may be relative concepts, which are used for relative description and clarification, and may change accordingly according to changes in the orientation of the components in the drawings.

[0049] Example 1

[0050] like Figure 1 As shown, an embodiment of the present invention provides an intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning, comprising:

[0051] The benchmark model generation subsystem is configured to acquire a continuous state stream (including time-series data such as leaf reflectance characteristics and branch morphology fluctuations) during the crop growth cycle through a multispectral imaging array. This continuous state stream is fed into a growth rhythm decoupler that separates environmental disturbances from intrinsic biological changes, extracting the ideal morphological benchmark vector for the crop at the current growth stage. The real-time image of the continuous state stream is then compared with the ideal morphological benchmark vector for a micro-variation field, outputting a biological stress difference map (retaining only features that deviate from the normal growth rhythm).

[0052] The mutually exclusive association engine subsystem is configured to synchronously input the biological stress difference map with environmental sensor data such as soil ion permeability and canopy microclimate flow. It then removes the noise of environmental factors in the difference map through the environmental interference stripping layer to generate a purified pathological response map. The purified pathological response map is then input into the pathological feature dissociation network, which outputs an environmental immune disease marker set (containing only abnormal features that cannot be explained by environmental factors).

[0053] The pathogen trajectory deduction subsystem is configured to input the disease marker set into the pathogen transmission dynamics model constructed by the aerosol deposition / root contact probability network of the pathogen in the crop population, and calculate the initial infection focus coordinates; based on the initial infection focus coordinates, combined with the real-time wind direction vector field and plant density topology map, through the biological diffusion dynamic chain.

[0054] In the above-mentioned embodiment, the benchmark model generation subsystem of this embodiment constructs a dynamic benchmark for the ideal growth state of crops through multispectral time series data streams, thereby removing environmental noise from growth anomaly detection; the micro-variation field comparison technology upgrades traditional static image recognition to growth rhythm deviation detection, thereby increasing the sensitivity of early disease and insect pest response to the subclinical stage. The mutually exclusive association engine subsystem eliminates false positives caused by abiotic stresses such as meteorology and soil through multimodal mutually exclusive analysis of environmental sensor data and biological stress characteristics; the environmental immune disease marker set output by the pathological feature dissociation network upgrades traditional disease identification to specific diagnosis at the etiology level. The pathogen trajectory deduction subsystem converts discrete disease markers into spatial epidemiological predictions based on the infection focus positioning of the transmission dynamics model; combining the diffusion dynamics chain of the microclimate field and plant topology, it realizes the deduction from single plant diagnosis to group transmission path, providing spatial coordinate guidance for precise prevention and control.

[0055] In summary, the three subsystems of this embodiment form a closed technical loop of "anomaly detection - cause diagnosis - spread prediction," upgrading traditional passive image recognition to an active monitoring system capable of tracing the cause of disease. A dual denoising mechanism, combining growth rhythm decoupling and environmental interference removal, addresses the high false positive rate in complex farmland scenarios. The pathogen transmission dynamics model transcends the limitations of existing technologies, which only achieve symptom classification, and outputs spatial control coordinates with agronomic value. Ultimately, this achieves a leap in monitoring from single-point identification to group-wide control.

[0056] Example 2

[0057] like Figure 2 As shown, based on Example 1, the benchmark model generation subsystem provided by the embodiment of the present invention includes:

[0058] The original state stream decomposition component is configured to input the continuous state stream collected by the multispectral imaging array into the growth rhythm decoupler, and extract the environment-independent feature set of the current growth stage through rhythm phase locking;

[0059] The dynamic synthesis component of the reference vector is configured as an environment-independent feature set through a morphological converger, which generates an ideal morphological reference vector that evolves over the growth cycle day by day. Its dimensions include biometric indicators such as leaf curvature gradient and branch fractal density.

[0060] The biorhythm deviation quantification component is configured to input the real-time image and the ideal morphological reference vector of the corresponding timestamp into the micro-variation field comparison unit, calculate the pixel-level biorhythm deviation through the rhythm offset kernel, and generate a biostress difference map that only retains the abnormal area.

[0061] In the aforementioned embodiment, the original state stream decomposition component of this embodiment uses a phase-locking algorithm of the growth rhythm decoupler to separate biological intrinsic features that are contaminated by environmental variables from multispectral time series data. This process upgrades traditional static spectral analysis to dynamic feature extraction that is adaptive to the growth stage, addressing the problem of feature drift caused by environmental noise. The dynamic reference vector synthesis component uses a morphological converger to perform nonlinear fitting of biometric indicators across days to generate ideal morphological reference vectors that continuously evolve over the growth cycle. This technology overcomes the limitations of fixed threshold detection, enabling time-varying modeling of crop growth states and providing a dynamic reference system for anomaly detection. The biorhythm deviation quantification component calculates the local deviation between real-time image pixels and the reference vector using the rhythm offset kernel function of the micro-variation field comparison unit. The biostress difference map output by this module uses spatial filtering to suppress normal growth areas, focusing detection on true anomaly sites.

[0062] In summary, the three components of this embodiment form a processing chain of "environmental interference removal - dynamic benchmark construction - anomaly quantification output", realizing the transformation from raw multispectral data to precise anomaly detection. Through the dynamic synthesis of time-series growth benchmarks and pixel-level deviation calculation, the traditional disease identification based on fixed thresholds is upgraded to an early anomaly warning system that adapts to growth rhythms, significantly improving the detection sensitivity of lesions. The construction of an environmentally independent feature set also ensures that the detection results are not interfered with by abiotic stress factors.

[0063] Example 3

[0064] Based on Example 2, the biorhythm deviation quantification component provided in this embodiment of the present invention includes:

[0065] The multi-dimensional rhythm space mapping subcomponent is configured to input real-time images and ideal morphological reference vectors into the rhythm offset kernel, and the real-time images are parsed into pixel-level biological rhythm vectors; the reference vectors are expanded into spatiotemporal reference fields;

[0066] The rhythm offset field calculation subcomponent is configured to calculate the spectral rhythm offset of each pixel by using a phase detuning detector; and calculate the morphological rhythm offset of each structure by using a morphological inertia operator;

[0067] The abnormal focus field synthesis subcomponent is configured to input spectral rhythm offset and morphological rhythm offset into the abnormal focus field generator, and fuse the two types of offsets into a comprehensive rhythm deviation field according to biological weights; the healthy rhythm fluctuations are filtered out through the rhythm coherence filter, and only the deviation signals that exceed the crop tolerance threshold are retained.

[0068] In the aforementioned embodiments, this embodiment first uses dynamic phase analysis to continuously track the spectral characteristics of crop growth. Phase demodulation techniques in signal processing are then used to detect the phase difference between real-time spectral data and the ideal growth model, thereby quantifying the degree of temporal deviation of the spectral characteristics. Simultaneously, morphological change trend modeling is based on the growth inertia characteristics of crop structure. By analyzing the dynamic evolution of morphological parameters such as branches and leaves, a time-varying morphological reference curve is established. Trend matching is calculated between the real-time observed morphological data and the reference curve to identify morphological variations that deviate from the normal growth trajectory. These two methods work together to capture abnormal fluctuations in spectral reflectance characteristics and identify unconventional changes in morphological development. Ultimately, multi-dimensional data fusion enables accurate quantification of rhythm deviations. Overall, this method achieves quantitative analysis of biorhythm deviations from the pixel level to the structural level. Through a cascade of three steps: spatiotemporal reference field mapping, multi-dimensional offset calculation, and abnormal signal focusing, a comprehensive rhythm deviation field is ultimately output that accurately reflects abnormal crop conditions.

[0069] Example 4

[0070] like Figure 3 As shown, based on Example 1, the mutually exclusive association engine subsystem provided by this embodiment of the present invention includes:

[0071] The environmental-pathological spectrum decoupling component is configured to input the biological stress difference map and environmental sensor data into the environmental interference stripping layer; the soil ion permeability and canopy microclimate flow through the environmental coupler to generate an environmental stress response spectrum; the biological stress difference map and the environmental stress response spectrum are separated by a frequency domain mutual exclusion filter to obtain a purified pathological response spectrum;

[0072] A multi-order pathology feature extraction component is configured to purify the pathology response map and input it into the pathology feature dissociation network to perform multi-order pathology feature extraction;

[0073] First-level dissociation: Filtering non-infectious anomalies such as mechanical damage through a spatial correlation filter (topological continuity analysis of lesion spread); Second-level dissociation: Separating pest and disease characteristics through a temporal mutation detector (differences in the time-frequency response of pest infestation and pathogen infection); Third-level dissociation: Using a spectral attenuation pattern recognizer (absorption attenuation fingerprints of fungal hyphae / bacterial secretions in specific bands) to identify pathogen types;

[0074] The marker set output component is configured to input the three-level dissociation result into the environmental immunity verification gate and output the environmental immunity type disease marker set.

[0075] In the above-mentioned embodiment, this embodiment achieves the precise separation of environmental stress and true pathological response in crop anomaly detection through the cascade processing of environmental interference stripping and multi-order pathological feature extraction. After the environmental-pathological spectrum decoupling component eliminates environmental noise interference, the multi-order pathological feature extraction component sequentially uses a triple filtering mechanism of spatial topology analysis, time-frequency response identification, and spectral fingerprint recognition to gradually eliminate non-infectious damage, distinguish between pests and diseases, and ultimately identify the pathogen type. The marker set output component verifies the output environmental immune disease marker set to form a pure pathological feature library that is not interfered with by environmental factors, providing specific indicators for accurate diagnosis. The entire process extracts biomarkers with pathological specificity from the mixed signals step by step.

[0076] Example 5

[0077] like Figure 4 As shown, based on Example 1, the pathogen trajectory deduction subsystem provided by this embodiment of the present invention includes:

[0078] The infection source probability field construction component is configured to input the symptom marker set into the pathogen transmission dynamics model; construct an airborne transmission trajectory mechanism based on the airborne transmission characteristics of pathogen spores, and construct an underground contact diffusion mechanism based on the expansion characteristics of soil pathogen hyphae; the airborne transmission trajectory mechanism and the underground contact diffusion mechanism are convolved and superimposed by the spatial symptom density and transmission probability to calculate the coordinates of the initial infection focus;

[0079] The three-dimensional biological diffusion simulation component is configured to input the biological diffusion dynamic chain with the focal coordinates, superimpose the real-time wind direction vector field through the turbulence corrector to generate the air transmission probability flow; superimpose the plant density topology map through the underground resistance estimator to generate the soil contact probability flow; the air transmission probability flow and the soil contact probability flow are output through the cross-medium coupler to output the three-dimensional transmission probability cloud;

[0080] The risk thermal field conversion component is configured as a risk energy level converter that maps the energy of the crop resistance threshold and the pathogen outbreak critical point into a three-dimensional transmission probability cloud input; it marks the over-threshold area in the transmission probability cloud through tolerance critical detection; and generates a disease thermal threat field by weighting the over-threshold area through time domain attenuation.

[0081] In the aforementioned embodiments, this embodiment models the air / soil dual-channel transmission mechanism, integrating convolution operations between spatial disease density and transmission probability to accurately calculate the spatial coordinates of the initial infection source. Atmospheric dynamics (wind turbulence correction) and underground topological resistance are integrated to construct a dual-medium air / soil transmission probability flow, forming a three-dimensional pathogen diffusion probability cloud. Crop resistance thresholds and pathogen outbreak critical parameters are introduced to achieve a quantitative conversion of the biological diffusion model to a risk energy level, generating a thermal threat field with time-domain attenuation characteristics.

[0082] In summary, this implementation transforms the biological process of disease transmission into a quantifiable risk thermal field by coupling multiple physical fields (air / soil / biological resistance) and superimposing spatiotemporal dimensions (three-dimensional space + time-domain attenuation). This provides a spatially explicit decision-making basis for disease early warning. Each module forms a complete computational closed loop from pathogen tracing to risk prediction, and the output directly guides the demarcation of prevention and control zones and the selection of intervention opportunities.

[0083] Example 6

[0084] Based on Example 5, the infection source probability field construction component provided by the embodiment of the present invention includes:

[0085] The airborne transmission trajectory mechanism construction subcomponent is configured to input the disease marker set into an aerosol deposition network that includes the coupled adsorption effect of pathogen spore particle size and canopy turbulent kinetic energy; generate dynamic deposition trajectories based on the spore air suspension persistence; and quantify the deposition potential energy of spores at each plant site using vortex attachment nuclei.

[0086] The subcomponent constructs the underground contact diffusion mechanism, configured to use a set of disease markers to drive a root contact network that topologically matches the pressure of pathogen hyphae tips with cracks in the root epidermis. The optimal hyphae path set is generated based on the root interlacing density of the overlapping rhizospheres of adjacent plants. The contact potential energy is calculated based on the biochemical reaction rate of hyphae secreting extracellular enzymes to dissolve root waxes.

[0087] The infection focus convolution aggregation sub-component is configured as an infection potential energy aggregator with a dual-mechanism output and input field theory superposition of spatial disease density and propagation potential energy; the sedimentation potential energy and contact potential energy are coupled through the cross-medium potential field to generate a comprehensive infection potential field; the comprehensive infection potential field is located by the extreme value of the three-dimensional potential field gradient change rate, and the initial infection focus coordinates are output.

[0088] In the above-mentioned embodiments, the airborne propagation trajectory modeling of this embodiment constructs an aerosol deposition network based on the coupled adsorption effect of spore particle size and canopy turbulence, generates a dynamic deposition trajectory in combination with the spore suspension persistence, and quantifies the deposition potential energy of the plant site through vortex attachment nuclei. Underground contact diffusion modeling constructs a root contact network based on the matching relationship between hyphae tip pressure and root fissures, calculates the optimal hyphae path set based on rhizosphere overlap, and quantifies the contact potential energy through the extracellular enzyme dissolution rate. Infection focus positioning integrates air deposition potential energy and underground contact potential energy to form a comprehensive infection potential field, adopts field theory superposition method for cross-medium potential field coupling, and determines the initial infection focus coordinates based on the extreme value of the three-dimensional potential field gradient.

[0089] In summary, this example, through coupled calculations of aerosol dynamics and mycelial growth biomechanics, transforms the physical and biochemical processes of pathogen transmission into a quantifiable infection potential field, ultimately achieving precise spatial localization of the initial infection source. This component provides high-precision input coordinates for the subsequent diffusion simulation module, forming the basis for calculating initial conditions for disease transmission simulation.

[0090] Example 7

[0091] Based on Example 6, the infection focus convolution aggregation subcomponent provided in the embodiment of the present invention includes:

[0092] A potential energy input module is configured as a cross-medium potential field coupler for inputting the sedimentation potential energy and the contact potential energy into the quantitative mechanism of the biochemical chain reaction of the airborne spore sedimentation triggering the secondary infection of the soil;

[0093] The biopotential field fusion module is configured to accelerate the expansion of hyphae through the biochemical pathway of secreting signal molecules through spore germination, converting sedimentation potential energy into a soil infection enhancement coefficient. The contact potential energy activates hyphae invasion through the rhizosphere, inducing crops to release volatile organic compounds that attract spores, and converts it into an airborne transmission enhancement coefficient. The dual enhancement coefficients and the original potential energy are superimposed on the biofield to form a three-dimensional comprehensive infection potential field.

[0094] The potential field gradient extreme value positioning module is configured as a potential energy peak detector based on the principle of critical energy transition of pathogen groups breaking through crop defenses in the three-dimensional comprehensive infection potential field input; the gradient change rate tensor of each point in the potential field space is calculated through the biopotential gradient mechanism of electrochemical potential transition at the junction of healthy tissue and infected tissue; the gradient change rate tensor is locked to the maximum infection potential energy point through the extreme value convergence kernel, and the output three-dimensional coordinates are the coordinates of the initial infection focus.

[0095] In the aforementioned embodiment, the infection focus convolution aggregation sub-component system of this embodiment achieves dynamic quantitative modeling and spatial localization of the airborne spore cross-medium infection process through the synergistic action of multiple modules. Its core significance can be broken down into three levels: The cross-medium energy conversion mechanism. The potential energy input module constructs an energy transfer framework across the three interfaces of "air-soil-plant," quantifying the airflow-conducted sedimentation potential energy (spore deposition dynamics) and contact potential energy (spore attachment efficiency) into calculable physical parameters, providing an energy input benchmark for subsequent biological processes. The biofield coupling enhancement system. The biopotential field fusion module establishes a unique dual-coefficient conversion mechanism: sedimentation potential energy is converted into a soil infection enhancement coefficient through the biochemical pathway of spore germination → hyphal expansion; contact potential energy is converted into an airborne transmission enhancement coefficient through the biochemical chain of rhizosphere activation → VOC release. These two factors are vectorially superimposed with the original potential energy in the three-dimensional biofield to form a comprehensive infection potential field that incorporates the characteristics of medium interaction. The mathematical representation of the critical infection point and the potential field gradient extreme value positioning module apply differential geometry methods to calculate the gradient change rate tensor of the potential field and combine it with the biological boundary conditions of the electrochemical potential transition; so that the biological process of pathogen groups breaking through the host defense is converted into a quantifiable energy topological structure.

[0096] In summary, this example constructs a complete analytical chain of "physical energy input - biological process conversion - mathematical extreme value location," providing a computable physical-biological model for predicting the initial infection site of trans-medium pathogen transmission. Its output (the coordinates of the initial infection focus) is essentially a spatial interpretation of the dynamic balance between interfacial energy transfer, biochemical transformation, and critical host defense conditions.

[0097] Example 8

[0098] Based on Example 7, the potential field gradient extreme value positioning module provided in the embodiment of the present invention includes:

[0099] The infection electrochemical field construction submodule is configured to input the three-dimensional integrated infection potential field into the biopotential gradient operator that responds to the opening and closing of ion channels in the plant cell membrane before and after pathogen invasion. The infection potential field is converted into a tissue potential field through a mechanism that quantifies the transmembrane potential difference between healthy and diseased cells. An ion flux detection core, which follows the diffusion trajectory of pathogen-secreted effector proteins that disrupt cellular ion balance, is used to mark the electrochemical transition interface in the potential field.

[0100] The gradient tensor dynamic generation submodule is configured as a potential differential unit for the spatial conductivity characteristics of ion concentration gradients within the plant vascular system, which is input into the electrochemical transition interface; the plasmodesmata potential conductivity is calculated along the leaf vein vessel direction to determine the anisotropic resistance of cell wall pores to ion flow; and the xylem potential decay rate is calculated along the stem vascular bundle direction to determine the turbulent diffusion effect of pathogenic toxins within the vessel.

[0101] The full-space tensor synthesis submodule is configured as a biofield tensor fusion device that fuses the double differential results through the topological covariance of the electrochemical response of plant tissue to generate a gradient change rate tensor.

[0102] In the above-mentioned embodiment, the potential field gradient extreme value location module of this embodiment achieves quantitative characterization of the dynamic changes in the potential field of plant tissues during pathogen infection through the synergistic effect of three submodules: electrochemical field construction, dynamic gradient tensor generation, and full-space tensor synthesis. The infection electrochemical field construction submodule converts the three-dimensional infection potential field into a tissue potential field and marks the electrochemical transition interface. The dynamic gradient tensor generation submodule calculates the potential conductivity of the plasmodesmata along the veins and the potential decay rate of the xylem along the stem based on the conductive properties of the plant vascular system. The full-space tensor synthesis submodule integrates the double differential results to generate a gradient rate of change tensor.

[0103] In summary, this embodiment maps the biological infection process into electrochemical field changes through a transmembrane potential difference quantification mechanism. A dynamic gradient model of the potential field is established by combining the conductive characteristics of the plant vascular system. Tensor fusion is used to comprehensively characterize the local electrophysiological perturbations and global diffusion effects at the infection front. This provides a differentiable mathematical framework for the spatiotemporal evolution of pathogen infection, supporting the localization and prediction of infection foci.

[0104] Example 9

[0105] Based on Example 8, the full-space tensor synthesis submodule provided by the embodiment of the present invention includes:

[0106] The conductivity orthogonal base construction unit is configured to input the plasmodesmata potential conductivity and the xylem potential attenuation rate into the tissue conductivity orthogonalization unit of the conductivity anisotropy of the spongy tissue and palisade tissue of the plant leaf; the plasmodesmata potential conductivity is calibrated by the leaf vein admittance calibrator based on the spatial modulation effect of the ion flow on the micropore distribution of the mesophyll cell wall, thereby generating a transverse conductivity weight field;

[0107] The three-dimensional covariance computation unit is configured to perform conductance-attenuation coupling at the vascular bundle intersection node, the ion redistribution hub at the petiole-stem junction. It calculates the tensor covariance coefficients using the nodal flow conservation law for charge conservation in local crop tissues under pathogen infection.

[0108] The global tensor generation unit is configured as a biofield tensor synthesis kernel of the tensor covariance coefficient through the covariance derivative rule of the electrochemical parameters of the plant tissue in the Euclidean space, and outputs the gradient change rate tensor.

[0109] In the aforementioned embodiment, this embodiment achieves spatial decoupling of plant tissue conductivity parameters through orthogonal basis building blocks. A vein admittance calibrator establishes a quantitative mapping between ion flow and tissue structure (micropore distribution, plasmodesmata), and a weighted field generator converts anisotropic conductivity properties into computable scalar parameters. A three-dimensional covariant computation unit implements the following at the vascular bundle node: nonlinear coupling of conductivity and potential decay rate, maintenance of transtissue current continuity based on charge conservation, and dynamic parameter correction under pathogenic interference conditions. A biofield tensor synthesis kernel is used to achieve: differential geometric expression of tissue electrochemical parameters, biological realization of covariant derivatives in Euclidean space, and multiscale integration of gradient rate tensors.

[0110] In summary, this embodiment establishes an analytical relationship between tissue structural parameters and electrical conductivity characteristics, realizes dynamic coupling calculation under pathological / physiological conditions, and outputs an interpretable bioelectric activity tensor field.

[0111] Example 10

[0112] Based on Example 9, the three-dimensional covariant computing unit provided in this embodiment of the present invention includes:

[0113] The node ion flow balance detection subunit is configured as a node flow conservation detector based on the mass conservation law of bidirectional ion flow at healthy vascular bundle nodes to obtain a theoretical ion flux balance value; and a pathogenic ion interference library based on the leakage rate database of ion channels blocked by specific pathogenic effector proteins is called to generate actual ion leakage.

[0114] The charge conservation constraint solver is configured to input the theoretical ion flux balance value and the actual ion leakage into the conservation constraint solver. The minimum leakage compensation flow is obtained through the ion compensation mechanism from adjacent healthy tissue to the infected node. The minimum leakage compensation flow is converted into a node covariant scaling factor through the infection-compensation balance.

[0115] The global covariance coefficient synthesis subunit is configured as a tensor covariance mapper that inputs the node covariance scale factor into the covariance transmission rule determined by the topological structure of the vascular network; the spatial weight coefficient of each node is obtained according to the vascular bundle hierarchical conduction law (ion flow attenuation gradient from main stem → branch → leaf vein); the spatial weight coefficient and the node covariance scale factor are generated into a tensor covariance coefficient through a hierarchical covariance aggregator.

[0116] In the aforementioned embodiment, the node flow conservation detector of this embodiment calculates the theoretical ion flux balance value of a healthy vascular node based on the law of conservation of mass. The pathogenic ion interference library provides ion channel leakage rate data caused by pathogenic effector proteins to calculate the actual ion leakage. The conservation constraint solver compares the theoretical ion flux balance value with the actual ion leakage value, calculates the minimum leakage compensation flow through the compensation mechanism of healthy tissue, and converts the infection-compensation balance into a node covariance scaling factor. The tensor covariance mapper calculates the spatial weight coefficient based on the covariance rule of the vascular network topology and the hierarchical conduction law (the attenuation gradient from main stem to branch to leaf vein). The hierarchical covariance aggregator integrates the spatial weight coefficient with the node covariance scaling factor to output the tensor covariance coefficient.

[0117] In summary, this embodiment constitutes a computational framework for the electrophysiological response of plant vascular networks under pathogen infection. Its core functions are to quantify the dynamic balance of ion flow under pathogen interference, calculate the compensation mechanism of healthy tissue for diseased nodes, and output tensor covariance coefficients that conform to the conduction characteristics of the vascular hierarchy.

[0118] Figure 5 A block diagram is shown of an exemplary electronic device that is suitable for implementing embodiments of the present invention.

[0119] The electronic device may include a central processing unit / microprocessor / main control chip, etc.; a storage medium, coupled to the central processing unit / microprocessor / main control chip, etc., and storing computer-executable instructions therein, for performing the steps of each method of an embodiment of the present invention when executed by the processor.

[0120] The central processing unit / microprocessor / main control chip etc. may include but is not limited to, for example, one or more processors or microprocessors etc.

[0121] The storage medium may include, but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disk, floppy disk, solid-state drive, removable disk, CD-ROM, DVD-ROM, Blu-ray disc, etc.).

[0122] In addition, the electronic device may also include (but not limited to) a data bus, an input / output bus / external bus / device bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.).

[0123] The central processing unit / microprocessor / main control chip etc. can communicate with external devices via an I / O bus via a wired or wireless network (not shown).

[0124] The storage medium may also store at least one computer-executable instruction for executing the various functions and / or method steps in the embodiments described in this technology when executed by a central processing unit / microprocessor / main control chip, etc.

[0125] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.

[0126] Figure 6 A schematic diagram of a computer-readable storage medium according to an embodiment of the present invention is shown.

[0127] like Figure 6 As shown, a non-transitory computer-readable storage medium stores instructions, such as computer-readable instructions. When the computer-readable instructions are executed by a processor, the various methods described above can be executed. Non-transitory computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-transitory non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0128] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0129] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0130] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0131] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the various embodiments of the method of the present invention via a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0132] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning, characterized in that: Include: The pathogen trajectory deduction subsystem is configured to input the disease marker set into a pathogen transmission dynamics model constructed by the aerosol deposition and root contact probability network of the pathogen in the crop population, and calculate the coordinates of the initial infection focus. Based on the initial infection focus coordinates, combined with the real-time wind direction vector field and plant density topology map, the biological diffusion dynamic chain is used to deduce the transmission path from single plant diagnosis to the population. The pathogen trajectory deduction subsystem includes: The infection source probability field construction component is configured to input the symptom marker set into the pathogen transmission dynamics model; construct an airborne transmission trajectory mechanism based on the airborne transmission characteristics of pathogen spores, and construct an underground contact diffusion mechanism based on the expansion characteristics of soil pathogen hyphae; the airborne transmission trajectory mechanism and the underground contact diffusion mechanism are convolved and superimposed by the spatial symptom density and transmission probability to calculate the coordinates of the initial infection focus; The three-dimensional biological diffusion deduction component is configured to input the biological diffusion dynamic chain with the focus coordinates, superimpose the real-time wind direction vector field through the turbulence corrector, and generate the airborne transmission probability flow; The superimposed plant density topology map is passed through the underground resistance evaluator to generate a soil contact probability flow; the air transmission probability flow and the soil contact probability flow are output through the cross-medium coupler to output a three-dimensional transmission probability cloud; The risk thermal field conversion component is configured as a risk energy level converter that uses a three-dimensional propagation probability cloud as input to map the energy between crop resistance thresholds and pathogen outbreak critical points; The over-threshold areas in the propagation probability cloud are marked by tolerance critical detection; the over-threshold areas are weighted by time domain attenuation to generate the disease thermal threat field.

2. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 1, characterized in that: The infection source probability field construction components include: The airborne transmission trajectory mechanism construction subcomponent is configured to input the disease marker set into an aerosol deposition network that includes the coupled adsorption effect of pathogen spore particle size and canopy turbulent kinetic energy; generate dynamic deposition trajectories based on the spore air suspension persistence; and quantify the deposition potential energy of spores at each plant site using vortex attachment nuclei. The subcomponent constructs the underground contact diffusion mechanism, configured to use a set of disease markers to drive a root contact network that topologically matches the pressure of pathogen hyphae tips with cracks in the root epidermis. The optimal hyphae path set is generated based on the root interlacing density of the overlapping rhizospheres of adjacent plants. The contact potential energy is calculated based on the biochemical reaction rate of hyphae secreting extracellular enzymes to dissolve root waxes. The infection focus convolution aggregation sub-component is configured as an infection potential energy aggregator with a dual-mechanism output and input field theory superposition of spatial disease density and propagation potential energy; the sedimentation potential energy and contact potential energy are coupled through the cross-medium potential field to generate a comprehensive infection potential field; the comprehensive infection potential field is located by the extreme value of the three-dimensional potential field gradient change rate, and the initial infection focus coordinates are output.

3. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 2, characterized in that: Infection focus convolution aggregation subcomponent, including: A potential energy input module is configured as a cross-medium potential field coupler for inputting the sedimentation potential energy and the contact potential energy into the quantitative mechanism of the biochemical chain reaction of the airborne spore sedimentation triggering the secondary infection of the soil; The biopotential field fusion module is configured to accelerate the expansion of hyphae through the biochemical pathway of secreting signal molecules through spore germination, converting sedimentation potential energy into a soil infection enhancement coefficient. The contact potential energy activates hyphae invasion through the rhizosphere, inducing crops to release volatile organic compounds that attract spores, and converts it into an airborne transmission enhancement coefficient. The dual enhancement coefficients and the original potential energy are superimposed on the biofield to form a three-dimensional comprehensive infection potential field. The potential field gradient extreme value positioning module is configured as a potential energy peak detector based on the principle of critical energy transition of pathogen groups breaking through crop defenses in the three-dimensional comprehensive infection potential field input; the gradient change rate tensor of each point in the potential field space is calculated through the biopotential gradient mechanism of electrochemical potential transition at the junction of healthy tissue and infected tissue; the gradient change rate tensor is locked to the maximum infection potential energy point through the extreme value convergence kernel, and the output three-dimensional coordinates are the coordinates of the initial infection focus.

4. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 3, characterized in that: Potential field gradient extreme value positioning module, including: The infection electrochemical field construction submodule is configured to input the three-dimensional integrated infection potential field into the biopotential gradient operator that responds to the opening and closing of ion channels in the plant cell membrane before and after pathogen invasion. The infection potential field is converted into a tissue potential field through a mechanism that quantifies the transmembrane potential difference between healthy and diseased cells. An ion flux detection core, which follows the diffusion trajectory of pathogen-secreted effector proteins that disrupt cellular ion balance, is used to mark the electrochemical transition interface in the potential field. The gradient tensor dynamic generation submodule is configured as a potential differential unit for the spatial conductivity characteristics of ion concentration gradients within the plant vascular system, which is input into the electrochemical transition interface; the plasmodesmata potential conductivity is calculated along the leaf vein vessel direction to determine the anisotropic resistance of cell wall pores to ion flow; and the xylem potential decay rate is calculated along the stem vascular bundle direction to determine the turbulent diffusion effect of pathogenic toxins within the vessel. The full-space tensor synthesis submodule is configured as a biofield tensor fusion device that fuses the double differential results through the topological covariance of the electrochemical response of plant tissue to generate a gradient change rate tensor.

5. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 4, characterized in that: Full-space tensor synthesis submodule, including: The conductivity orthogonal base construction unit is configured to input the plasmodesmata potential conductivity and the xylem potential attenuation rate into the tissue conductivity orthogonalization unit of the conductivity anisotropy of the spongy tissue and palisade tissue of the plant leaf; the plasmodesmata potential conductivity is calibrated by the leaf vein admittance calibrator based on the spatial modulation effect of the ion flow on the micropore distribution of the mesophyll cell wall, thereby generating a transverse conductivity weight field; The three-dimensional covariance computation unit is configured to perform conductance-attenuation coupling at the vascular bundle intersection node, the ion redistribution hub at the petiole-stem junction. It calculates the tensor covariance coefficients using the nodal flow conservation law for charge conservation in local crop tissues under pathogen infection. The global tensor generation unit is configured as a biofield tensor synthesis kernel of the tensor covariance coefficient through the covariance derivative rule of the electrochemical parameters of the plant tissue in the Euclidean space, and outputs the gradient change rate tensor.

6. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 5, characterized in that: Three-dimensional covariant computing unit, including: a node ion flow balance detection subunit configured to obtain a theoretical ion flux balance value through a node flow conservation detector based on the mass conservation law of bidirectional ion flow at a healthy vascular bundle node; The pathogenic ion interference library of the leakage rate database of ion channels blocked by specific pathogenic effector proteins is called to generate the actual ion leakage amount; The charge conservation constraint solver is configured to input the theoretical ion flux balance value and the actual ion leakage into the conservation constraint solver. The minimum leakage compensation flow is obtained through the ion compensation mechanism from adjacent healthy tissue to the infected node. The minimum leakage compensation flow is converted into a node covariant scaling factor through the infection-compensation balance. The global covariance coefficient synthesis subunit is configured as a tensor covariance mapper that inputs the node covariance scale factor into the covariance transmission rule determined by the vascular network topology structure; the spatial weight coefficient of each node is obtained according to the vascular bundle hierarchical transmission law; the spatial weight coefficient and the node covariance scale factor are generated into a tensor covariance coefficient through a hierarchical covariance aggregator.

7. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 6, characterized in that: The vascular bundle level conduction law in the global covariance coefficient synthesis subunit is the ion flow attenuation gradient from the main stem to the branches and from the branches to the leaf veins.

8. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 1, characterized in that: It also includes: a benchmark model generation subsystem, which is configured to obtain a continuous state stream containing leaf reflection characteristics and branch morphology fluctuation time series data during the crop growth cycle through a multispectral imaging array; the continuous state stream is input into a growth rhythm decoupler that separates environmental interference from biological intrinsic changes, and extracts the ideal morphological benchmark vector of the crop in the current growth stage; the real-time image of the continuous state stream is compared with the ideal morphological benchmark vector for a micro-change field, and a biological stress difference map that only retains the characteristics that deviate from the normal growth rhythm is output.

9. The intelligent monitoring system for crop diseases and insect pests based on image recognition and machine learning according to claim 8, characterized in that: It also includes: a mutually exclusive association engine subsystem, which is configured to synchronously input the biological stress difference map with the soil ion permeability and canopy microclimate flow environmental sensor data; through the environmental interference stripping layer, the noise of environmental factors in the difference map is eliminated to generate a purified pathological response map; the purified pathological response map is input into the pathological feature dissociation network, and an environmental immune disease marker set is output that only contains abnormal features that cannot be explained by environmental factors.

Images