Method and system for monitoring and predicting plant diseases and insect pests of greenhouse vegetables based on Internet of Things

Abstract

The invention relates to the technical field of smart agriculture and agricultural Internet of Things, and discloses a method and system for monitoring and predicting facility vegetable diseases and insect pests based on Internet of Things, and the method comprises the steps: firstly, measuring the heat plume transmission delay time; an active photo-thermal excitation unit is controlled to emit a modulation light signal, heat wave propagation is generated in the blade, and directional thermally induced convection plume is induced; setting a holographic capture phase window based on the delay time, and cooperatively acquiring a blade surface infrared image sequence and an aerosol particle diffraction pattern; thermal impedance characteristics reflecting stomatal conductance are extracted through digital phase locking operation, and pathogen concentration is identified through holographic reconstruction; and finally, fusing thermal impedance and pathogen concentration information to carry out multi-dimensional judgment. According to the method, the physiological status and pathogen information of the plant are synchronously obtained by utilizing a heat flow coupling mechanism, non-infectious physiological stress and early disease infection are effectively distinguished, and accurate and active monitoring and early warning of the plant diseases and insect pests of the greenhouse vegetables are realized.

Description

Technical Field

[0001] This invention relates to the fields of smart agriculture and agricultural Internet of Things (IoT) technology, specifically to a method and system for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things. Background Technology

[0002] Currently, greenhouse vegetable cultivation, as an important industry for ensuring a stable supply of agricultural products, is expanding its production scale. However, greenhouses generally possess microclimate characteristics of high temperature, high humidity, and relatively closed airflow, which easily induces various fungal diseases and pests. If timely intervention is not carried out in the early stages of disease outbreaks, it often leads to a decline in crop yield and quality. Therefore, utilizing modern Internet of Things (IoT) technology for real-time and continuous monitoring of crop growth environment and health status has become a key link in achieving precision management in greenhouse agriculture and reducing pesticide use.

[0003] For disease monitoring in greenhouse environments, existing technologies mainly employ monitoring schemes based on machine vision image recognition or environmental spore capture. Machine vision schemes typically involve distributing visible light or multispectral cameras within the greenhouse to periodically acquire images of the crop canopy. Algorithms are then used to analyze changes in color and texture on the leaf surface to identify lesions. Spore capture schemes, on the other hand, often utilize inhalation samplers. Built-in fans draw in air from the greenhouse, causing airborne particles to collide and settle onto glass slides or adhesive tape. A microscopic imaging system is then used to identify and count the morphological characteristics of the settled spores, thereby inferring the probability of disease occurrence.

[0004] While the aforementioned methods have achieved some degree of automated monitoring, several technical bottlenecks remain. First, visual representation-based methods suffer from latency, typically only identifying lesions or necrotic tissue visible to the naked eye on leaf surfaces. By this time, pathogens have often already infected and spread throughout the plant, missing the optimal control window during the incubation period. Second, traditional spore-catching devices often employ a wide-area passive aspiration mode, lacking spatial directionality in sampling. They cannot distinguish whether captured spores originate from the target plant being monitored or from environmental background noise drifting with the airflow, and are easily affected by dust in greenhouses, leading to a high false alarm rate. Furthermore, in existing monitoring systems, thermal infrared information reflecting the internal physiological state of plants and spore information reflecting the presence of external pathogens are often acquired independently, lacking physical and temporal correlation. Single thermal imaging data struggles to differentiate between physiological stomatal closure caused by drought and defensive stomatal closure caused by disease, resulting in the system's inability to accurately identify non-infectious physiological stress and early risks of infectious diseases.

[0005] Therefore, this invention provides a method and system for monitoring and predicting diseases and pests in facility vegetables based on the Internet of Things, in order to address the shortcomings of existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and system for monitoring and predicting diseases and pests in greenhouse vegetables based on the Internet of Things (IoT). This solves the problems of existing monitoring technologies for diseases and pests in greenhouse vegetables, such as limited monitoring dimensions, difficulty in distinguishing between physiological stress and early disease infection, and delayed early warning.

[0007] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things, comprising the following steps: S1. Obtain the physical distance parameters between the monitoring terminal and the target blade, and between the target blade and the air intake of the monitoring terminal, and collect the current ambient temperature and relative humidity. Based on this, calculate the thermal plume transmission delay time required for the thermally induced convection plume to travel from the surface of the target blade to the air intake. S2. Control the active photothermal excitation unit to emit modulated periodic light signals to the target blade, generate heat wave propagation inside the target blade, and heat the air boundary layer on the surface of the target blade to induce a directional flow of thermally induced convection plume. S3. The infrared image sequence of the target blade surface is acquired by the infrared thermal imaging sensing unit. At the same time, the holographic capture phase window is calculated according to the thermal plume transmission delay time. The microfluidic lensless holographic sensing unit is triggered to acquire the diffraction pattern of aerosol particles only when the phase of the periodic light signal is within the holographic capture phase window. S4. Perform digital phase-locked loop operation on the infrared image sequence, extract the thermal wave signal component with the same frequency as the periodic light signal, calculate the phase hysteresis map of the target blade surface, and extract the thermal impedance characteristics reflecting the porosity based on the phase hysteresis map. S5. Perform numerical backpropagation reconstruction on the diffraction pattern to obtain a complex amplitude distribution image of the particles, extract the morphological feature parameters of the particles from the complex amplitude distribution image and compare them with the pathogen feature database to obtain the concentration value of the target pathogen. S6. Determine the thermal impedance abnormality level based on the thermal impedance characteristics, determine the spore concentration abnormality level based on the concentration value of the target pathogen, and combine the thermal impedance abnormality level and the spore concentration abnormality level according to the preset fusion logic rules to generate an early warning level and output control commands.

[0008] By adopting the above technical solution, this invention utilizes active photothermal excitation to synchronously drive thermal wave signals and thermo-induced convection. Through digital phase-locked thermal imaging technology and time-series synchronous holographic sensing technology, it non-destructively acquires thermal impedance characteristics reflecting the physiological state of plants and pathogen particle information in the environment. Based on dual-dimensional data fusion judgment, it achieves early and accurate identification and differentiated early warning of diseases.

[0009] Preferably, the process of calculating the thermal plume transport delay time in step S1 includes: determining the air density, aerodynamic viscosity, and air thermal expansion coefficient based on the current ambient temperature and relative humidity; calculating the average rising velocity of the thermally induced convective plume based on the average temperature rise of the target blade surface caused by the periodic light signal and the air thermal expansion coefficient; and dividing the vertical distance from the center of the target blade surface to the air inlet of the microfluidic lensless holographic sensing unit by the average rising velocity to obtain the thermal plume transport delay time.

[0010] By adopting the above technical solution and calibrating the delay time using a fluid dynamics model, the acquisition time of the holographic sensing unit is precisely synchronized with the arrival time of the pathogen particles, thereby improving the signal-to-noise ratio.

[0011] Preferably, the process of calculating the holographic capture phase window in step S3 includes: converting the thermal plume transmission delay time into a phase delay angle according to the modulation frequency of the periodic optical signal; superimposing the phase delay angle on the energy peak phase of the periodic optical signal to obtain a center phase; setting a phase interval of a preset width before and after the center phase, and using the phase interval as the holographic capture phase window.

[0012] By adopting the above technical solution, a correlation between time and phase was established, enabling precise timing control of the acquisition action and effectively filtering out background particle interference not originating from the target blade.

[0013] Preferably, the process of calculating the phase lag map of the target blade surface in step S4 includes: for each pixel in the infrared image sequence, extracting the temperature value change data of the pixel over time to form a temperature time series; performing discrete Fourier transform or integration operations on the temperature time series with a preset in-phase and in-frequency reference signal and a quadrature reference signal to obtain the in-phase component and quadrature component of the pixel temperature response signal at the modulation frequency; determining the phase lag value of the pixel by performing an arctangent operation on the ratio of the quadrature component to the in-phase component, and constructing the phase lag map from the phase lag values ​​of all pixels.

[0014] By adopting the above technical solution, a weak phase hysteresis signal is extracted from a noisy thermal image sequence using a digital phase-locked loop algorithm. This signal is directly related to the blade thermal impedance and can sensitively reflect changes caused by early stress responses such as stomatal closure.

[0015] Preferably, the process of combining and judging according to the preset fusion logic rules in step S6 includes: when the abnormal thermal impedance level reaches a preset threshold and the abnormal spore concentration level is normal, it is determined to be a non-infectious physiological stress state; when the abnormal spore concentration level reaches a preset threshold and the abnormal thermal impedance level is normal, it is determined to be a potential risk state of environmental pathogens; when the abnormal thermal impedance level and the abnormal spore concentration level both reach a preset first-level threshold, it is determined to be an early infection risk state of the disease.

[0016] By adopting the above technical solution and using multi-dimensional information fusion for judgment, it is possible to effectively distinguish between physiological stress caused by environmental factors such as water shortage and stress caused by pathogen infection, thus avoiding false alarms and achieving accurate diagnosis.

[0017] Secondly, the present invention provides an Internet of Things-based monitoring and prediction system for pests and diseases in facility vegetables, for performing the aforementioned method, including a heat flow coupling monitoring terminal, a network transmission module, and an application layer server; The thermal flow coupling monitoring terminal includes an edge computing and collaborative control unit, an active photothermal excitation unit, a microfluidic lensless holographic sensing unit, and an infrared thermal imaging sensing unit. The active photothermal excitation unit is installed at the lower part of the thermal-fluid coupling monitoring terminal and is used to irradiate the selected area from below the target blade and induce an upward thermally induced convection plume. The air intake of the microfluidic lensless holographic sensing unit is installed on the upper part of the thermal-fluid coupling monitoring terminal and is aligned directly above the irradiation area of ​​the active photothermal excitation unit, for receiving the thermally induced convection plume and collecting the diffraction pattern of the particles carried therein. The infrared thermal imaging sensing unit is installed on the thermal flow coupling monitoring terminal, and its field of view covers the irradiation area of ​​the active photothermal excitation unit, for acquiring infrared image sequences. The edge computing and collaborative control unit is connected to the active photothermal excitation unit, the microfluidic lensless holographic sensing unit, and the infrared thermal imaging sensing unit, respectively, and is used to generate modulation signals, calculate holographic capture phase windows, and issue synchronous acquisition pulses.

[0018] By adopting the above technical solution, this system achieves the collaborative operation of thermal imaging sensing and holographic aerosol sensing at the hardware level through an integrated thermal flux coupling monitoring terminal. It has a compact structure, is easy to deploy, and can stably perform dual-modal information synchronous acquisition tasks, providing a reliable hardware foundation for achieving accurate monitoring and prediction.

[0019] This invention provides a method and system for monitoring and predicting pests and diseases in greenhouse vegetables based on the Internet of Things (IoT). It has the following beneficial effects: 1. This invention employs a combination of active photothermal excitation and thermo-induced convection plume induction. It utilizes modulated light signals to generate heat waves propagating within the leaf while simultaneously heating the leaf surface to induce a directional thermo-induced convection plume. This heat-fluid coupling mechanism not only enables non-destructive detection of internal physiological characteristics such as stomatal conductance through heat wave signals, but also utilizes the thermal plume as a physical carrier to actively transport pathogen particles from the leaf surface to the sensing interface. This overcomes the shortcomings of traditional passive sedimentation sampling methods, such as low efficiency and high randomness, and achieves synchronous and active monitoring of plant microscopic physiological states and surface pathogen information.

[0020] 2. This invention calculates the thermal plume transport delay time based on environmental calibration parameters and establishes a holographic capture phase window accordingly. The microfluidic lensless holographic sensing unit is triggered only within a specific phase interval of the excitation signal for data acquisition. This phase gating mechanism achieves precise targeting of aerosol particles originating from the target leaf surface in the time domain. It effectively filters out background dust or irrelevant spores randomly floating in the greenhouse environment that do not originate from the currently monitored target, improving the signal-to-noise ratio and source certainty of pathogen detection.

[0021] 3. This invention constructs a multidimensional spatiotemporal fusion decision logic based on thermal impedance characteristics and pathogen concentration values, comprehensively analyzing phase lag information reflecting plant physiological functions and spore count information reflecting environmental biological stress. This logic can accurately distinguish between non-infectious physiological stress caused by water and nutrient deficiencies and infectious diseases caused by pathogen invasion. Furthermore, it can identify early risks during the incubation period before symptoms appear through dual-indicator anomalies, solving the problem that a single monitoring dimension is insufficient for qualitatively distinguishing stress types. This enables precise diagnosis and differentiated grading early warning of diseases and pests in greenhouse vegetables. Attached Figure Description

[0022] Figure 1 This is an architecture diagram of the IoT-based facility vegetable pest and disease monitoring and prediction system of the present invention; Figure 2 This is a flowchart of the IoT-based monitoring and prediction method for pests and diseases in facility vegetables according to the present invention. Detailed Implementation

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] See attached document Figure 1The present invention provides an Internet of Things-based monitoring and prediction system for pests and diseases in facility vegetables. The system includes: at least one heat flow coupling monitoring terminal, a network transmission module, and an application layer server.

[0025] This system adopts a three-layer IoT architecture. The perception layer consists of multiple heat flow coupling monitoring terminals deployed within the greenhouse facility, responsible for the collection and front-end processing of raw data. The network layer consists of wireless transmission modules, using technologies such as LoRa, ZigBee, NB-IoT, or Wi-Fi, to transmit the data collected by the perception layer to the application layer. The application layer consists of edge computing gateways or cloud servers, responsible for in-depth analysis, fusion judgment, model storage and updates of all collected data, and generating early warning commands.

[0026] The heat flow coupling monitoring terminals are deployed in the greenhouse using either a matrix layout or a layout based on key monitoring points. In a matrix layout, the terminals are evenly distributed with preset row and column spacing to cover the entire planting area. In a layout based on key monitoring points, terminals are preferentially deployed in areas with a history of high disease incidence, near greenhouse vents, or in areas with significant changes in environmental temperature and humidity gradients.

[0027] The physical layout of the thermal-fluid coupling monitoring terminal forms the material basis for achieving the technical effects of this invention. Each monitoring terminal integrates an active photothermal excitation unit, a microfluidic lensless holographic sensing unit, an infrared thermal imaging sensing unit, and an edge computing and collaborative control unit. These units have specific spatial relative positions.

[0028] An active photothermal excitation unit is installed at the lower part of the monitoring terminal, enabling it to irradiate selected leaf areas from the side or directly below the canopy of the target vegetable plant. The physical layout of the microfluidic lensless holographic sensing terminal forms the material basis for achieving the technical effects of this invention. Each monitoring terminal integrates an active photothermal excitation unit, a microfluidic lensless holographic sensing unit, an infrared thermal imaging sensing unit, and an edge computing and collaborative control unit. These units have specific spatial relative positions.

[0029] The active photothermal excitation unit is installed at the bottom of the monitoring terminal, enabling it to irradiate selected leaf areas from the side or directly below the canopy of the target vegetable plant. The air intake of the microfluidic lensless holographic sensing unit is installed at the top of the monitoring terminal, precisely aligned above the leaf area irradiated by the active photothermal excitation unit, maintaining a vertical distance of 5 to 15 centimeters from the leaf surface. The infrared thermal imaging sensing unit is installed at a top-down angle of 45 to 90 degrees to ensure that its field of view completely covers the excited leaf area.

[0030] This spatial arrangement forms a physical closed loop. The energy applied to the blade by the active photothermal excitation unit not only serves as the heat source for infrared thermal imaging sensing, but also induces an upward thermo-convective plume on the blade surface. The path of this thermo-convective plume points precisely towards the intake port of the microfluidic lensless holographic sensing unit. Therefore, the excitation process directly drives the sampling process, constituting a physical linkage between excitation and sampling.

[0031] The active photothermal excitation unit includes a high-power light-emitting diode (LED) array and a modulation controller. The LED array uses a near-infrared light source, such as LEDs with wavelengths in the 850-940 nm range, to ensure that its radiant energy is effectively absorbed by plant leaves and converted into heat energy, while avoiding significant interference with plant photosynthesis. The modulation controller is a programmable circuit used to generate a drive current with a specific frequency and waveform (e.g., a sine wave or square wave) according to instructions from the edge computing and collaborative control unit, to control the periodically modulated radiant power output of the LED array.

[0032] The microfluidic lensless holographic sensing unit comprises a miniature air pump, a transparent substrate with microfluidic channels, a monochromatic coherent light source, and a CMOS image sensor. The miniature air pump generates negative pressure, drawing air into the microfluidic channels through an air intake. The inner walls of the microfluidic channels are transparent, allowing light to pass through. The monochromatic coherent light source, such as a 635 nm laser diode, is positioned on one side of the microfluidic channels. The CMOS image sensor is placed directly on the other side of the microfluidic channels, opposite the laser diode, without any optical lenses between them. As the laser beam passes through aerosol particles within the microfluidic channels, diffraction occurs, and the diffraction pattern is directly recorded by the CMOS image sensor.

[0033] The infrared thermal imaging sensing unit uses an uncooled microbolometer focal plane array as its core detector, with a spectral response range covering the long-wave infrared band from 8 to 14 micrometers. The frame rate of this unit is set to be at least 10 times higher than the modulation frequency of the active photothermal excitation unit to ensure accurate capture of periodic temperature changes caused by thermal excitation. For example, if the modulation frequency is 0.1 Hz, the acquisition frame rate of the infrared thermal imaging sensing unit is no less than 1 Hz.

[0034] The edge computing and collaborative control unit is the control core of the monitoring terminal, typically implemented by a system-on-a-chip (SoC) integrating a field-programmable gate array (FPGA) and a microcontroller (MCU). The MCU is responsible for communicating with the application layer server and managing task scheduling. The FPGA is responsible for high-precision timing control, generating modulation signals to drive the active photothermal excitation unit, and precisely generating synchronous acquisition pulses to trigger the infrared thermal imaging sensing unit and the microfluidic lensless holographic sensing unit based on the phase reference of the modulation signal, thereby achieving strict synchronization and phase locking of all sensing units in the time domain.

[0035] See attached document Figure 2 This invention provides a method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things, which is executed by the system in Example 1.

[0036] First, step S1 is executed to perform environmental calibration and transmission delay measurement. In this step, the edge computing and collaborative control unit acquires the physical distance between the active photothermal excitation unit of the monitoring terminal and the target blade, as well as the physical distance between the target blade and the air intake of the microfluidic lensless holographic sensing unit. Simultaneously, the system collects the current ambient temperature and humidity parameters and calculates the time required for the thermally induced convective plume to travel from the blade surface to the air intake based on an aerodynamic model; this is the thermal plume transmission delay time, providing a benchmark for subsequent timing control.

[0037] Next, step S2 is executed to implement periodic thermal excitation and thermal plume induction. In this step, the system controls the active photothermal excitation unit to emit a periodic light signal modulated by amplitude or frequency towards the target blade. This light signal is absorbed by the blade and converted into heat energy, generating periodic heat waves that propagate within the blade and simultaneously heat the air boundary layer on the blade surface. The heated air density decreases and rises, forming a stable thermo-convective plume carrying microscopic particles from the blade surface. This plume flows directionally towards the intake of the microfluidic lensless holographic sensing unit.

[0038] Subsequently, step S3 is executed, performing collaborative data acquisition based on a phase gating mechanism. In this step, the infrared thermal imaging sensing unit continuously acquires a sequence of infrared images of the blade surface to record the spatiotemporal changes in temperature. Simultaneously, the edge computing and collaborative control unit calculates a specific holographic capture phase window based on the transmission delay time determined in step S1. Only when the phase of the excitation signal is within this window is the system triggered by the microfluidic lensless holographic sensing unit to acquire the diffraction pattern of aerosol particles, thereby ensuring that the acquired particles mainly originate from the excited blade surface and effectively filtering out irrelevant environmental background noise.

[0039] Next, step S4 is executed to perform phase-locked thermal wave signal processing and feature extraction. In this step, the system performs digital phase-locked operation on the acquired time-domain infrared image sequence. By extracting the thermal wave signal component with the same frequency as the excitation frequency, the amplitude map and phase hysteresis map of the blade surface are calculated. This phase hysteresis map can eliminate the interference of surface emissivity non-uniformity and light reflection, and highly sensitively characterize the thermal impedance distribution of the blade, thereby reflecting physiological pathological features such as abnormal stomatal conductance or early water stress.

[0040] Next, step S5 is executed to complete the holographic reverse reconstruction and pathogen identification. In this step, the system preprocesses the acquired lensless diffraction pattern of aerosol particles to remove background noise. Then, a numerical backpropagation algorithm is applied to reconstruct the diffraction pattern into a complex amplitude distribution image of the particles. The system further extracts the morphological features of the particles from the reconstructed image and compares them with a pre-set pathogen database, thereby achieving automatic identification and counting of specific pathogen spores.

[0041] Finally, step S6 is executed to implement multi-dimensional spatiotemporal fusion judgment and early warning. In this step, the system logically fuses the phase lag information reflecting the physiological state obtained in step S4 with the spore count results reflecting the presence of pathogens obtained in step S5. Based on the distribution of abnormal thermal impedance areas and the threshold determination of pathogen concentration, the system distinguishes between different states such as non-infectious physiological stress, disease incubation period, and disease outbreak period, generates corresponding early warning levels, and outputs control commands.

[0042] The monitoring and prediction method of the present invention further includes step S1, namely, environmental calibration and transmission delay measurement. This step is performed when the monitoring terminal is first deployed or when the system is restarted, providing the necessary timing parameters for subsequent synchronous data acquisition.

[0043] During system initialization, the edge computing and collaborative control unit first reads and confirms the fixed physical parameters stored in its internal non-volatile memory. These parameters include the vertical distance from the center of the light spot of the active photothermal excitation unit to the surface of the target blade, and the vertical distance from the center of the target blade surface to the air intake of the microfluidic lensless holographic sensing unit. Simultaneously, environmental sensors integrated into the monitoring terminal, such as digital temperature and humidity sensors, are activated to measure the air temperature and relative humidity around the current monitoring point.

[0044] After obtaining these basic parameters, the edge computing and collaborative control unit performs the calculation of the thermal plume transport delay time. This calculation is based on a simplified fluid dynamics model that treats thermal convection as a natural convection process in still air. The edge computing and collaborative control unit, based on the measured air temperature and relative humidity, queries its internally stored lookup tables or calls built-in functions to obtain the key physical properties of the air under these environmental conditions, including air density, aerodynamic viscosity, and the coefficient of thermal expansion.

[0045] Based on the above parameters, using a pre-defined calculation model, such as by solving simplified fluid dynamics equations or applying empirical formulas, the updraft velocity induced on the blade surface by the slight temperature rise generated by active photothermal excitation is calculated. Subsequently, by dividing the vertical distance from the blade surface to the intake by the updraft velocity, the time required for the thermal plume carrying particles to travel from the blade surface to the intake is obtained; this time is the thermal plume transport delay time.

[0046] Calculate the thermal plume transport delay time The formula is: ; in, Indicates the thermal plume transport delay time (in seconds); This represents the vertical distance (in meters) from the center of the target blade surface to the intake port of the microfluidic lensless holographic sensing unit. This represents the average upward velocity of the thermally induced convection plume (m / s). This represents the acceleration due to gravity (taken as 9.8 m / s²). 2 ); This represents the coefficient of thermal expansion of air (1 / K). This indicates the average temperature rise of the blade surface relative to ambient air caused by active photothermal excitation. ).

[0047] The calculated thermal plume transport delay time is used as a key timing calibration parameter and temporarily stored in the dynamic random access memory of the edge computing and collaborative control unit for direct retrieval in step S3 when calculating the holographic capture phase window. This calibration process ensures the accuracy of the acquisition timing, enabling it to adapt to changes in installation height and ambient temperature and humidity.

[0048] After performing step S1, the monitoring and prediction method of the present invention proceeds to step S2, namely periodic thermal excitation and thermal plume induction. This step is the physical driving force for subsequent synchronous detection of physiological states and pathogens.

[0049] In this step, the edge computing and collaborative control unit controls the active photothermal excitation unit to emit periodically varying light radiation towards the selected target blade region. The intensity of this light radiation is controlled by a preset modulation signal, the waveform of which can be a sine wave or a square wave. The frequency of the modulation signal is set within a specific low-frequency range, typically 0.01 Hz to 0.5 Hz. This frequency range is chosen to ensure that the heat wave has sufficient time to diffuse within the blade tissue, thereby obtaining a considerable phase response, while avoiding excessively high frequencies that would cause the thermal response signal to decay too quickly.

[0050] The physical process is as follows: Near-infrared light emitted by the active photothermal excitation unit is absorbed by water and pigments in the leaf tissue. According to the photothermal effect, the absorbed light energy is converted into heat energy, causing the internal temperature of the leaf to rise and fall periodically. This temperature fluctuation forms a three-dimensional heat wave that propagates from the excitation center to the surrounding tissues.

[0051] Meanwhile, the periodic temperature changes on the blade surface directly heat the thin layer of air immediately adjacent to the blade surface, namely the air boundary layer. When the air within the boundary layer is heated, its density decreases. Under the influence of gravity, the less dense, hot air experiences buoyancy relative to the surrounding, more dense, cold air, thus moving upwards. Because the thermal excitation is continuous and periodic, this upward-moving air forms a stable, continuous updraft, namely a thermoconvective plume. The initial driving force of this plume comes from the temperature rise of the blade surface, and it carries particles attached to the blade surface (such as pathogenic spores, dust, etc.) upwards along a vertical path.

[0052] After step S2 is executed, the monitoring and prediction method of the present invention proceeds to step S3, namely, phase-gated collaborative data acquisition. This step utilizes precise timing control to achieve synchronous and highly correlated sampling of leaf physiological state and leaf surface pathogens.

[0053] In this step, data acquisition is performed collaboratively. First, the infrared thermal imaging sensing unit, triggered by the synchronization signal from the edge computing and collaborative control unit, continuously acquires infrared images of the target blade surface at a constant frame rate, generating an infrared image sequence containing at least one complete thermal excitation cycle. The acquisition rate of this sequence is higher than the excitation frequency to fully record the periodic response of the blade surface temperature to the excitation signal.

[0054] Simultaneously, the edge computing and collaborative control unit performs the calculation of the holographic capture phase window. It first calculates the period of the excitation signal based on the modulation frequency used by the active photothermal excitation unit in step S2. Then, it converts the thermal plume transport delay time calibrated in step S1 into a phase delay. This phase delay represents the phase angle shift required from when the blade surface temperature reaches its peak until the airflow carrying that heat and particles reaches the intake port of the microfluidic lensless holographic sensing unit.

[0055] The system superimposes this phase delay onto the peak phase of the thermal excitation signal and sets an extremely narrow time window before and after this central phase; this window is the holographic capture phase window. For example, for a sinusoidal excitation signal, its energy input peak is at a 90-degree phase. The system adds the phase angle corresponding to the transmission delay to 90 degrees and sets a phase interval of 5 to 10 degrees as the capture window, centered on this.

[0056] Only when the real-time phase of the excitation signal enters this pre-calculated holographic capture phase window does the edge computing and collaborative control unit issue a trigger pulse, instructing the microfluidic lensless holographic sensing unit to acquire a diffraction pattern. This process is called phase gating. Since the generation of the thermal plume is directly related to the increase in blade surface temperature, and the increase in blade surface temperature is closely related to the peak energy input of the thermal excitation, sampling within a specific phase window that is synchronized with the peak thermal excitation and takes into account the transmission delay ensures that the vast majority of the collected aerosol particles originate from the group that has just been stripped from the blade surface by thermal convection and transported up. In this way, the system physically filters out background particles that are randomly floating in the environment at other times and are unrelated to the currently monitored target blade, thereby improving the signal-to-noise ratio and source certainty of pathogen detection.

[0057] After collaborative data acquisition is completed, the monitoring and prediction method of this invention proceeds to step S4, namely phase-locked thermal wave signal processing and feature extraction. This step aims to extract feature images from the time-domain infrared image sequence that can accurately reflect the physiological state of the leaves.

[0058] In this step, the edge computing and collaborative control unit processes the infrared image sequence acquired in step S3, which covers at least one complete excitation cycle. The core of the processing is a digital lock-in amplification algorithm. For each pixel in the infrared image sequence, the system treats its temperature value over time as an independent signal.

[0059] The system performs correlation calculations on the temperature time series of the pixel with a reference signal that is in phase and frequency with the excitation signal in step S2, and another reference signal that is orthogonal to the same frequency. Specifically, the real and imaginary parts of the Fourier coefficients of the pixel's temperature response signal at the excitation frequency are calculated through discrete Fourier transform or direct integration.

[0060] Then, the amplitude and phase of the temperature response at that pixel are calculated based on these two components. The amplitude represents the severity of the temperature fluctuation at that point, while the phase represents the delay of the temperature response relative to the excitation signal. By repeating this process for each pixel in the entire infrared image sequence, the system ultimately generates two feature images of the same size as the original infrared image: one is an amplitude map, and the other is a phase map.

[0061] For any pixel in the image The phase hysteresis value of its temperature response Calculate according to the following formula: ; in, Represents pixels The phase lag value (in radians) at the point directly characterizes the local thermal impedance characteristics of the blade; and These represent the in-phase and quadrature components of the temperature response signal at the excitation frequency, respectively. Represents pixels In the The original temperature value of the frame; Indicates the total number of frames sampled; Indicates the modulation frequency (Hz) of the active photothermal excitation; This indicates the sampling frame rate (FPS) of the infrared thermal imaging sensor unit.

[0062] The phase map is a key output of this step. The pixel values ​​in the phase map directly reflect the delay experienced by heat waves as they propagate within the leaf. This delay is directly related to the thermophysical parameters of the leaf tissue, particularly thermal diffusivity and heat capacity. Plant leaves dissipate heat through stomatal transpiration. When stomatal conductance decreases (e.g., during water stress or early disease stages), heat dissipation efficiency decreases, local thermal resistance increases, and the speed of heat wave propagation slows down, resulting in a greater phase delay on the phase map. Therefore, the phase map provides spatial distribution information about stomatal activity and water status within the leaf.

[0063] Compared to directly using temperature or amplitude maps, phase maps inherently possess robustness against interference. Since the phase is derived by calculating the temporal relationship between the response and excitation signals, it is insensitive to static or slowly varying noise such as uneven emissivity on the blade surface, non-uniform illumination, and slow drift in ambient temperature. This allows phase map-based analysis to detect subtle changes in thermal properties caused by physiological stress earlier and more stably.

[0064] Parallel to or following step S4, the monitoring and prediction method of the present invention performs step S5, namely, holographic reverse reconstruction and pathogen identification. The purpose of this step is to convert the lensless diffraction pattern acquired in step S3 into an identifiable particle image and to complete the classification and counting of pathogens.

[0065] The first step is to preprocess the acquired raw lensless diffraction pattern. Preprocessing includes two main operations: First, background subtraction, which subtracts a pre-acquired background diffraction pattern free of particles from the currently acquired diffraction pattern to eliminate static interference fringes caused by inhomogeneities of the coherent light source, fixed-mode noise of the CMOS sensor, and surface defects of the microfluidic chip. Second, normalization, which linearly scales the intensity values ​​of the processed diffraction pattern to a standard range to compensate for minor fluctuations in light source intensity at different acquisition times.

[0066] After preprocessing, the edge computing and collaborative control unit or the upper-layer server executes a numerical backpropagation algorithm to reconstruct the preprocessed diffraction pattern. This embodiment uses the angular spectral propagation algorithm. This algorithm simulates the diffraction process of light waves in the frequency domain: first, a Fourier transform is performed on the two-dimensional diffraction pattern to convert it to the spatial frequency domain; then, a phase transfer function is multiplied in the frequency domain, which describes the phase change of the light wave as it propagates backward from the CMOS sensor plane to the original object plane where the particle is located; finally, an inverse Fourier transform is performed on the product result to computationally reconstruct the complex amplitude distribution image of the plane where the particle is located. This complex amplitude image simultaneously contains the particle's morphological information (amplitude image) and phase delay information (phase image).

[0067] After obtaining the reconstructed complex amplitude image, the system performs segmentation processing to identify individual particle targets within the image. For each segmented particle target, the system calculates a series of morphological feature parameters. These parameters include the particle's equivalent diameter, aspect ratio, roundness, area, perimeter, and internal texture features.

[0068] Finally, the system compares the extracted morphological feature parameter vector of each particle with a pathogen feature database pre-stored in a local or cloud database. This database contains the standard morphological feature parameter ranges of spores of various target diseases (such as downy mildew and powdery mildew). Using a classifier algorithm, such as support vector machine or nearest neighbor classification, the system categorizes each particle into a specific pathogen type or a non-pathogen particle. After classification, the system counts the pathogen particles of a specific type to obtain the concentration value of the target pathogen in this sampling.

[0069] After completing steps S4 and S5, the monitoring and prediction method of the present invention finally executes step S6, namely, multi-dimensional spatiotemporal fusion decision and early warning. This step integrates data from two independent but physically related measurement dimensions to generate a comprehensive diagnosis and early warning instruction for plant health status.

[0070] In this step, the edge computing and collaborative control unit or application layer server first analyzes the phase map generated in step S4 to determine abnormal thermal impedance. The system calculates the average phase value of all pixels in the phase map and compares it with the stored baseline phase value representing the health status. When the average phase value exceeds a preset primary physiological stress threshold, the system determines that the plant is under mild physiological stress. When the average phase value further exceeds a higher secondary physiological stress threshold, the system determines that the plant is under significant physiological stress.

[0071] Simultaneously, the system analyzes the pathogen concentration values ​​obtained in step S5 to determine abnormal spore concentrations. The system compares the concentration value of a specific pathogen type with a preset primary threshold. This threshold represents the background safe concentration of that pathogen spores in the environment. When the concentration value is below this threshold, the spore concentration is considered normal. When the concentration value exceeds the primary threshold but is below a higher secondary threshold, the spore concentration is determined to be abnormal. When the concentration value exceeds the secondary threshold, the spore concentration is determined to be severely excessive.

[0072] The system combines the results of abnormal thermal impedance and abnormal spore concentration into a single judgment based on preset fusion logic rules. The specific logical branches are as follows: If only thermal impedance abnormality reaches the first or second-level threshold, while spore concentration is normal, the system determines that the plant is under non-infectious physiological stress, such as simple water or nutrient stress. If only spore concentration abnormality reaches the first-level threshold, while thermal impedance is normal, the system determines it as a potential risk from environmental pathogens, but the plant has not yet shown stress symptoms. If both thermal impedance and spore concentration abnormalities reach the first-level threshold, the system determines it as an early-stage disease infection risk. If thermal impedance abnormality reaches the second-level threshold, while spore concentration abnormalities reach either the first or second-level threshold, the system determines it as an outbreak of infectious disease.

[0073] Based on the final judgment, the system outputs warning signals with clearly defined levels. For example, non-infectious physiological stress is defined as Level 1 warning (blue), potential risks from environmental pathogens as Level 2 warning (yellow), early disease infection risk as Level 3 warning (orange), and outbreaks of infectious diseases as Level 4 warning (red). Each warning level corresponds to a set of preset linkage control strategies. For example, a Level 1 warning triggers the irrigation system; a Level 3 warning triggers the precision pesticide application system and notifies management personnel; and a Level 4 warning triggers the highest priority intervention measures. These warning signals and control commands are transmitted through the network layer to the application layer server and the corresponding environmental control actuators.

Claims

1. A method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things, characterized in that, Includes the following steps: S1. Obtain the physical distance parameters between the monitoring terminal and the target blade, and between the target blade and the air intake of the monitoring terminal, and collect the current ambient temperature and relative humidity. Calculate the thermal plume transmission delay time required for the thermally induced convection plume to travel from the surface of the target blade to the air intake based on the physical distance parameters, the current ambient temperature, and the relative humidity. S2. Control the active photothermal excitation unit to emit modulated periodic light signals to the target blade, generate heat wave propagation inside the target blade, and heat the air boundary layer on the surface of the target blade to generate a directional thermally induced convection plume. S3. The infrared image sequence of the target blade surface is acquired by the infrared thermal imaging sensing unit. At the same time, the holographic capture phase window is calculated according to the thermal plume transmission delay time. The microfluidic lensless holographic sensing unit is triggered to acquire the diffraction pattern of aerosol particles only when the phase of the periodic light signal is within the holographic capture phase window. S4. Perform digital phase-locked loop operation on the infrared image sequence, extract the thermal wave signal component with the same frequency as the periodic light signal, calculate the phase hysteresis map of the target blade surface, and extract the thermal impedance characteristics reflecting the porosity based on the phase hysteresis map. S5. Perform numerical backpropagation reconstruction on the diffraction pattern to obtain a complex amplitude distribution image of the particles, extract the morphological feature parameters of the particles from the complex amplitude distribution image and compare them with the pathogen feature database to obtain the concentration value of the target pathogen. S6. Determine the thermal impedance abnormality level based on the thermal impedance characteristics, determine the spore concentration abnormality level based on the concentration value of the target pathogen, and combine the thermal impedance abnormality level and the spore concentration abnormality level according to the preset fusion logic rules to generate an early warning level and output control commands.

2. The method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things according to claim 1, characterized in that, The process of calculating the thermal plume transport delay time in step S1 includes: Determine the air density, aerodynamic viscosity, and air thermal expansion coefficient based on the current ambient temperature and relative humidity. The average upward velocity of the thermally induced convective plume is calculated based on the average temperature rise on the surface of the target blade caused by the periodic light signal and the coefficient of thermal expansion of the air. The thermal plume transport delay time is obtained by dividing the vertical distance from the center of the target blade surface to the air inlet of the microfluidic lensless holographic sensing unit by the average rising velocity.

3. The method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things according to claim 1, characterized in that, In step S2, the modulation waveform of the periodic optical signal is a sine wave or a square wave, the modulation frequency range of the periodic optical signal is set to 0.01 Hz to 0.5 Hz, and the active photothermal excitation unit uses a near-infrared light source with a wavelength in the range of 850 nm to 940 nm.

4. The method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things according to claim 1, characterized in that, The process of calculating the holographic capture phase window in step S3 includes: The thermal plume propagation delay time is converted into a phase delay angle based on the modulation frequency of the periodic optical signal; The phase delay angle is superimposed on the energy peak phase of the periodic optical signal to obtain the center phase; A phase interval of preset width is set before and after the center phase, and the phase interval is used as the holographic capture phase window. The microfluidic lensless holographic sensing unit only performs the acquisition action within the holographic capture phase window, filtering out environmental background noise that does not originate from the surface of the target blade.

5. The method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things according to claim 1, characterized in that, The process of calculating the phase hysteresis map of the target blade surface in step S4 includes: For each pixel in the infrared image sequence, the temperature value of the pixel is extracted over time to form a temperature time series. By performing discrete Fourier transform or integration operations on the temperature time series with a preset in-phase and in-frequency reference signal and a quadrature reference signal, the in-phase and quadrature components of the pixel temperature response signal at the modulation frequency are obtained. The phase lag value of the pixel is determined by performing an arctangent operation on the ratio of the quadrature component to the in-phase component. The phase lag map is constructed from the phase lag values ​​of all pixels, and the phase lag value is used to characterize the local thermal impedance characteristics of the target blade.

6. The method for monitoring and predicting diseases and pests in facility vegetables based on the Internet of Things according to claim 1, characterized in that, The process of numerical backpropagation reconstruction of the diffraction pattern in step S5 includes: The preprocessed diffraction pattern is then transformed into the spatial spectral domain by performing a two-dimensional Fourier transform. Multiply by a phase transfer function in the spatial spectrum domain, the phase transfer function describing the phase change of the light wave as it propagates backward from the sensor plane to the original object plane where the particle is located; A two-dimensional inverse Fourier transform is performed on the result of the multiplication to obtain the complex amplitude distribution image containing particle morphology information and phase information.

7. The method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things according to claim 1, characterized in that, The process of making a combination decision based on preset fusion logic rules in step S6 includes: When the abnormal thermal impedance level reaches a preset threshold and the abnormal spore concentration level is normal, it is determined to be a non-infectious physiological stress state. When the abnormal spore concentration level reaches a preset threshold and the abnormal thermal resistance level is normal, it is determined to be a potential risk state of environmental pathogens. When the abnormal thermal resistance level and the abnormal spore concentration level both reach the preset first-level threshold, it is determined to be a state of early disease infection risk. When the abnormal thermal impedance level reaches the preset second-level threshold and the abnormal spore concentration level reaches the preset first-level threshold or second-level threshold, it is determined to be an outbreak of infectious disease.

8. The method for monitoring and predicting diseases and pests in facility vegetables based on the Internet of Things according to claim 7, characterized in that, The control commands include: In response to the aforementioned non-invasive physiological stress state, a first-level early warning signal is output and the irrigation system is triggered; A secondary early warning signal is output based on the potential risk status of the environmental pathogens. In response to the early infection risk status of the disease, a three-level early warning signal is output and the precision application system is triggered; In response to the outbreak of the infectious disease, a level four early warning signal is output and the highest priority intervention measures are triggered.

9. The method for monitoring and predicting diseases and pests in facility vegetables based on the Internet of Things according to claim 6, characterized in that, Before performing numerical backpropagation reconstruction on the diffraction pattern, step S5 also includes a preprocessing step: Subtract the pre-collected background diffraction pattern (which does not contain particles) from the collected diffraction pattern to eliminate static interference fringe noise; The intensity values ​​of the diffraction pattern after background subtraction are linearly scaled to a preset standard range for normalization.

10. A monitoring and prediction system for pests and diseases in facility vegetables based on the Internet of Things, characterized in that, The method for monitoring and predicting pests and diseases in facility vegetables based on the Internet of Things as described in any one of claims 1 to 9 includes a heat flow coupling monitoring terminal, a network transmission module, and an application layer server. The thermal flow coupling monitoring terminal includes an edge computing and collaborative control unit, an active photothermal excitation unit, a microfluidic lensless holographic sensing unit, and an infrared thermal imaging sensing unit. The active photothermal excitation unit is installed at the lower part of the thermal-fluid coupling monitoring terminal and is used to irradiate the selected area from below the target blade and induce an upward thermally induced convection plume. The air intake of the microfluidic lensless holographic sensing unit is installed on the upper part of the thermal-fluid coupling monitoring terminal and is aligned directly above the irradiation area of ​​the active photothermal excitation unit, for receiving the thermally induced convection plume and collecting the diffraction pattern of the particles carried therein. The infrared thermal imaging sensing unit is installed on the thermal flow coupling monitoring terminal, and its field of view covers the irradiation area of ​​the active photothermal excitation unit, for acquiring infrared image sequences. The edge computing and collaborative control unit is connected to the active photothermal excitation unit, the microfluidic lensless holographic sensing unit, and the infrared thermal imaging sensing unit, respectively, and is used to generate modulation signals, calculate holographic capture phase windows, and issue synchronous acquisition pulses.

Images