Dynamic spectrum light supplementing device and method based on plant physiological signal feedback

Abstract

The invention provides a dynamic spectrum light supplementing device and method based on plant physiological signal feedback. The method comprises the following steps that at least one physiological optical signal of a light receiving plant is collected through a non-contact sensing unit; inputting the physiological optical signal into a preset plant physiological status diagnosis model, and outputting a current physiological status judgment result of the light-receiving plant; on the basis of the current physiological state judgment result, according to a predefined spectrum regulation and control strategy, generating a spectrum regulation and control instruction containing a light quality ratio parameter; and according to the spectrum regulation and control instruction, driving a multi-spectrum LED unit capable of independently dimming, and carrying out coordinated regulation and control on the plant canopy microclimate so as to carry out dynamic light supplement on the light receiving plant. According to the invention, the physiological status of the plant can be directly monitored, and dynamic and accurate spectrum regulation and control can be carried out accordingly.

Description

Technical Field

[0001] This invention relates to a dynamic spectral supplementary lighting device and method based on plant physiological signal feedback. Background Technology

[0002] In facility agriculture (such as greenhouses and plant factories), artificial lighting is a key technology for ensuring high-yield, high-quality, and high-efficiency crop production. Traditional or existing artificial lighting technologies mainly fall into the following categories: First, the most common method is fixed-spectrum supplemental lighting. This method uses a preset, fixed spectral ratio (such as a fixed red-blue light ratio), light intensity, and photoperiod to irradiate plants. Its advantages lie in its simple system structure and low cost. However, the light requirements of plants change dynamically at different growth and development stages (such as seedling stage, vegetative growth stage, and flowering and fruiting stage) and are affected by daily environmental fluctuations. A fixed-spectrum supplemental lighting strategy cannot respond to these changes, not only wasting energy but also failing to achieve the desired supplemental lighting effect, and may even inhibit plant growth due to an unsuitable light environment. Second, to further improve energy efficiency and adaptability, environmental factor feedback-based supplemental lighting has emerged. These systems monitor ambient light intensity using quantum sensors, automatically turning on the supplemental lights when the intensity falls below a set threshold and turning them off when the threshold is reached. Some improved systems can also adjust the supplemental light intensity within a certain range based on daytime or greenhouse temperature. Although this method is an improvement over fixed-spectrum supplemental lighting, it is still essentially "environment-driven," its regulation based on the physical environmental parameters within the greenhouse rather than the physiological state of the plants themselves. Plants' absorption and utilization of light involves complex physiological and biochemical processes, and their photosynthetic efficiency and physiological health cannot be accurately reflected by a single ambient light intensity. Therefore, environment-driven supplemental lighting cannot accurately determine whether a plant is suffering from light stress (such as photoinhibition or insufficient light), making it difficult to achieve truly on-demand supplemental lighting. Furthermore, to address the different needs of plants at different growth stages, pre-programmed spectral supplemental lighting technology has been proposed. This system pre-programs spectral programs for different crops at various growth stages and can automatically switch spectra according to a preset schedule. For example, it provides a spectrum that promotes stem and leaf growth during the vegetative growth stage and switches to a spectrum that promotes flowering during the flowering stage. While this method takes into account the plant's growth patterns, its switching is mechanical and time-triggered, failing to sense and respond to the real-time physiological state of the individual plant. When a plant's physiological state deviates from the preset growth model due to disease, nutrient deficiency, or other abiotic stresses, this rigid supplemental lighting pattern may not be effective and may even exacerbate the plant's physiological stress.

[0003] In summary, existing supplemental lighting technologies all suffer from a fundamental limitation: they fail to establish a real-time feedback control loop with the plant itself as the direct information source. Whether it's fixed spectrum, environmental feedback, or preset programs, none of them directly "listen" to the plant's own "needs" and cannot respond to the plant's instantaneous photosynthetic efficiency and physiological health. Therefore, there is an urgent need in this field for a novel supplemental lighting technology and device capable of directly monitoring the plant's physiological state and dynamically and precisely regulating the spectrum accordingly. Summary of the Invention

[0004] This invention provides a dynamic spectral supplementary lighting device and method based on plant physiological signal feedback, which can effectively solve the above-mentioned problems.

[0005] This invention is implemented as follows: A dynamic spectral supplemental lighting method based on plant physiological signal feedback includes the following steps: S1: Acquire at least one physiological optical signal from the plant receiving light through a non-contact sensing unit; S2: Input the physiological optical signal into a preset plant physiological state diagnosis model and output the current physiological state judgment result of the plant receiving light; S3: Based on the current physiological state judgment result, generate a spectral control instruction containing light quality ratio parameters according to a predefined spectral control strategy; S4: Based on the spectral control command, drive the independently dimmable multispectral LED unit and coordinate the microclimate of the plant canopy to provide dynamic supplemental lighting for the light-receiving plants.

[0006] A dynamic spectral supplemental lighting device based on plant physiological signal feedback, comprising: The sensing unit is configured for non-contact acquisition of physiological optical signals from light-receiving plants; The processing and control unit is communicatively connected to the sensing unit, and is used to receive the physiological optical signals, run the plant physiological state diagnostic model, and generate spectral regulation commands. The supplementary lighting and microclimate coordinated control unit is communicatively connected to the processing and control unit, and includes multiple independently controllable spectral channels and a microclimate adjustment mechanism for responding to the spectral control commands.

[0007] The beneficial effects of this invention are: (1) This invention collects plant physiological optical signals in a non-contact manner, judges its physiological state in real time based on an intelligent diagnostic model, and generates dynamic spectral control instructions based on photobiological principles. Finally, it drives the multispectral LED unit and the microclimate regulation mechanism to carry out coordinated regulation, thereby realizing a fundamental shift from "environment-driven" to "plant physiological-driven", significantly improving light energy utilization efficiency and effectively alleviating the adverse effects of environmental stress on plants.

[0008] (2) By deeply integrating spectral modulation and canopy microclimate modulation (fan) in hardware and execution logic, the system can simultaneously perform "optical load reduction" and "ventilation enhancement" when facing complex adversities such as high temperature stress. It exerts its power from three aspects: reducing heat load, strengthening evaporation cooling and ensuring CO2 supply. Its rate and effect of alleviating stress are significantly better than any single regulation method.

[0009] (3) The online transfer learning mechanism integrated in this invention enables the core of the system—the diagnostic model—to fine-tune and evolve itself using local data generated during operation. This gives the system personalized adaptation capabilities and allows it to continuously optimize to adapt to the unique characteristics of specific crop varieties, regional climates and greenhouse environments, thereby ensuring that it can always maintain optimal control performance in long-term operation.

[0010] (4) Based on the photochemical reflectance index (PRI) and the maximum photochemical quantum yield of PSII (Fv / Fm), two deep-seated mechanistic indicators that are extremely sensitive to the physiological state of plants, the system can provide early warning of physiological adversity before visible symptoms appear on the plant's surface. On this basis, the system optimizes the photosynthetic process of plants from a mechanistic perspective by performing functional spectral regulation based on photobiological principles (such as adjusting the proportion of far-red light to balance the energy of the photosystem) and achieves "precision phototherapy". Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0012] Figure 1 This is a system block diagram of the dynamic spectral supplementation device of the present invention.

[0013] Figure 2 This is a control flowchart of the dynamic spectral supplementation method of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. 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. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention.

[0015] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0016] Reference Figure 1-2 As shown, a dynamic spectral supplemental lighting method based on plant physiological signal feedback includes the following steps: S1: Acquire at least one physiological optical signal from the plant receiving light through a non-contact sensing unit; S2: Input the physiological optical signal into a preset plant physiological state diagnosis model and output the current physiological state judgment result of the plant receiving light; S3: Based on the current physiological state judgment result, generate a spectral control instruction containing light quality ratio parameters according to a predefined spectral control strategy; S4: Based on the spectral control command, drive the independently dimmable multispectral LED unit and coordinate the microclimate of the plant canopy to provide dynamic supplemental lighting for the light-receiving plants.

[0017] In step S1, the physiological optical signal includes the photochemical reflectance index obtained by a multispectral or hyperspectral imager, and / or the maximum photochemical quantum yield of PSII obtained by a chlorophyll fluorometer.

[0018] Specifically, step S1 is implemented as follows: S101. Employs a multispectral imaging camera with at least two core wavelengths of 531nm and 570nm, or a hyperspectral imager with higher spectral resolution. This module is installed 1.5 to 3 meters above the plant canopy and is responsible for acquiring the spectral reflectance information of the canopy over a wide area in a non-invasive manner. S102. A chlorophyll fluorometer with saturated flash pulse function is used, and an automatic dark adaptation leaf chamber is provided. The leaf chamber is made of opaque material and has a built-in miniature measurement probe for point-to-point and precise measurement of fluorescence parameters on selected representative leaves. S103. The two modules work together under the scheduling of the processing and control unit. The imaging module performs periodic wide-area scanning, while the fluorescence measurement module performs fine diagnosis of the leaves in key areas at specific times (such as early morning every day) or after receiving an abnormal trigger signal from the imaging module. S104 The imaging module is automatically triggered at a preset time interval (e.g., every 30 minutes). To ensure data accuracy, the acquisition process quickly completes image capture of the 531nm and 570nm bands within milliseconds, and simultaneously brings the diffuse reflection reference plate into the field of view for subsequent radiometric calibration. S105. The fluorescence measurement module follows a strict dark adaptation procedure. During planned measurements, its dark adaptation leaf chamber automatically closes, enveloping the leaf in complete darkness. After a precisely timed 20-30 minutes of dark adaptation, the module automatically executes the measurement sequence: first, it emits a weak measurement light to acquire the minimum fluorescence value F0, followed by the emission of a high-intensity saturated pulse of light (intensity > 3000 µmol m). - ² s - ¹, duration 0.5-1 second) to obtain the maximum fluorescence value Fm; S105 The processing and control unit performs radiometric calibration on the raw data acquired by the imaging module, converting the gray values ​​into physical reflectance. Then, it calculates the average PRI value of the canopy according to the formula PRI = (R531 - R570) / (R531 + R570). The processing and control unit receives the raw F0 and Fm data transmitted by the fluorescence measurement module and automatically calculates the maximum photochemical quantum yield of PSII according to the formula Fv / Fm = (Fm - F0) / Fm. Finally, the system adds a precise timestamp to each data point, and the continuous temporal trend of PRI and the fixed-point health baseline value of Fv / Fm are analyzed synchronously in the plant physiological state diagnostic model.

[0019] PRI (Photochemical Reflectance Index) is not simply an optical coefficient; it directly reflects the macroscopic optical properties of a plant's internal microscopic physiological and biochemical processes. Its core physiological basis is the xanthophyll cycle, the most important photoprotective mechanism in plants. When a plant absorbs more light energy than it can utilize for photosynthesis (i.e., light stress occurs), the excess light energy destroys the photosynthetic apparatus. At this point, the plant initiates the xanthophyll cycle, converting one type of xanthophyll (epoxyzeaxanthin) into another (zeaxanthin). The PRI (Photochemical Reflectance Index) comes from the fact that zeaxanthin can safely "quench" excess light energy through heat dissipation. This conversion process results in a very slight decrease in the reflectance of plant leaves near a wavelength of 531 nanometers (nm), because more light energy is used for non-photochemical quenching. The reflectance at a wavelength of 570 nm is relatively stable, as the formula PRI = (R531 - R570) / (R531 - R570) indicates this. The index, calculated by +R570, directly reflects the intensity of non-photochemical quenching by changes in its value (usually a decrease). Therefore, PRI is a sensitive indicator of whether the plant photosynthetic apparatus is in an "overloaded" state and whether its light energy utilization efficiency has decreased.

[0020] It should be noted that, unlike indices such as NDVI that reflect biomass accumulation, PRI can capture changes in plant physiological state at the minute to hour level. Before any visible symptoms (such as wilting or yellowing) appear on the plant's surface, a decrease in PRI can indicate a reduction in light energy utilization efficiency, providing a valuable early intervention window for the system. Simultaneously, it directly diagnoses the plant's "working state"—whether it is efficiently photosynthesizing or passively engaging in photoprotection. This allows the supplemental lighting method of this invention to be upgraded from "supplementing light quantity" to "optimizing light energy distribution." For example, when PRI decreases, this system does not simply increase the total light intensity (which may exacerbate stress), but rather promotes a balance of light energy between the two photosystems by adjusting the spectrum (such as increasing far-red light), fundamentally improving efficiency.

[0021] Specifically, in data acquisition, multispectral or hyperspectral imagers must be used because ordinary RGB cameras cannot capture the two specific and very close wavelengths of 531nm and 570nm. Imagers can provide spatial information and analyze the differences in photosynthetic efficiency in different parts of the canopy.

[0022] In step S2, the plant physiological state diagnostic model is a trained machine learning model, specifically implemented as follows: This invention employs an optimal machine learning model architecture to achieve accurate diagnosis for different types of sensor data. For time-series data (such as the continuous trend of PRI): Long Short-Term Memory (LSTM) network models are preferred. These models can effectively capture the long-term dependence of physiological signals over time, such as identifying predictive patterns like "PRI values ​​continuously decreasing over three consecutive monitoring periods," thus enabling early warning. For multidimensional instantaneous feature data (such as PRI, Fv / Fm, and canopy temperature from a single acquisition): Ensemble learning algorithms such as Random Forest or Gradient Boosting Machine can be used. These models can efficiently handle nonlinear relationships between features, have fast training speeds, and provide feature importance assessments. For spatial spectral data (such as multispectral images): Convolutional neural networks can be used to automatically extract the spatial distribution features of stress symptoms (such as the difference between the top and bottom of the canopy) from the images.

[0023] In step S4, the coordinated regulation of the plant canopy microclimate is achieved by adjusting the rotation speed of the fan integrated with the multispectral LED unit.

[0024] In step S4, the core driving part of the supplementary lighting and microclimate coordinated control unit consists of multiple independent constant current source circuits and a microcontroller. Each monochrome LED chip array (such as deep red, far red, and blue light arrays) is connected to an independent, programmable constant current driver chip (such as a professional LED driver IC based on TI's TLC5971). Similarly, the fan group is also connected to an independent motor drive circuit (such as a speed control circuit composed of MOSFETs). At the same time, the spectral control commands generated by the processing and control unit are sent to the supplementary lighting unit via a communication bus (such as I²C, SPI, or CAN bus). The MCU (Microcontroller Unit) incorporates a PWM (Pulse Width Modulation) generator. For each LED channel, the MCU generates a PWM square wave signal with a specific duty cycle based on the light quality ratio parameters specified in the instruction. This signal is then sent to the corresponding constant current driver chip. During the effective level of the PWM signal, the constant current driver chip outputs a constant current to illuminate the LED; during the ineffective level, it completely shuts off the current. By changing the proportion of time occupied by the effective level within a cycle (i.e., the duty cycle, typically expressed as 0% to 100%), stepless and linear adjustment of the LED's luminous intensity can be achieved. For example, if the instruction requires a 50% intensity for a deep red LED, the MCU outputs a PWM wave with a 50% duty cycle to the deep red LED's driver chip. By independently controlling the PWM duty cycle of all channels, the system can synthesize mixed light with arbitrary spectral power distributions within the range of 0% to 100%.

[0025] The multispectral LED module and the micro fan assembly are rigidly integrated into the same lamp housing. This not only simplifies wiring but, more importantly, ensures a high degree of spatial overlap between the illumination area and the airflow area. This allows the plant canopy to receive dual regulation from light and wind simultaneously and in the same location. At the software and communication protocol level, the control commands issued by the processing and control unit are a unified data packet for the entire supplemental lighting and microclimate coordinated control unit. This data packet contains both the PWM setting values ​​for each LED channel and the fan speed setting values. When the unit MCU receives this data packet, its firmware program updates all the new PWM values ​​and fan speed values ​​to the corresponding hardware registers almost synchronously within a very short control cycle (usually in milliseconds). This mechanism of "unified command packet" and "synchronous hardware update" ensures at the system level that there is no human-induced operational delay or communication sequence between the two actions of "lighting" and "winding." They are triggered simultaneously as an inseparable coordinated control event, ensuring that light and wind, as two environmental factors, can exert a common influence on the plant canopy at the same starting point in time. For example, when implementing strategies to combat high-temperature stress, plants do not first perceive a decrease in light intensity and then wait for a period of time before perceiving an increase in wind force; instead, they are instantly placed in a new, synergistically optimized complex environment of "decreased light intensity and increased wind force".

[0026] In step S3, the predefined spectral regulation strategy includes: when the diagnostic model determines that the plant is in a state of decreased photosynthetic efficiency, generating an instruction to increase the proportion of far-red light radiation in the spectrum; when the diagnostic model determines that the plant is in a state of high-temperature stress, generating an instruction to simultaneously reduce the proportion of red and far-red light and increase the fan speed. The specific implementation is as follows: Regulation of "decreased photosynthetic efficiency": When the plant physiological state diagnostic model outputs the state "decreased photosynthetic efficiency" and its confidence level is higher than a preset threshold (e.g., probability value > 0.7), this strategy is triggered. The system will determine a specific adjustment coefficient α (the value of α can be set from 0.1 to 0.5) from a predefined gain coefficient lookup table based on the decrease in PRI value relative to the healthy baseline. The spectral control command is specifically to increase the PWM (pulse width modulation) duty cycle of the far-red LED channel (700-750nm) to (1 + α) times the current value, while maintaining the duty cycle of other spectral channels unchanged, or adjusting it proportionally to maintain the relative stability of total photonic flux (PPFD). When photosynthetic efficiency declines (often accompanied by a decrease in PRI value), the excitation energy between photosystem II (PSII) and photosystem I (PSI) is often unbalanced. By precisely increasing the proportion of far-red light, PSI can be preferentially activated, thereby promoting the smooth operation of the electron transport chain and rebalancing the energy distribution between the two photosystems. This effectively alleviates photoinhibition, promotes the relaxation of the lutein cycle, directs more light energy towards photosynthetic carbon assimilation, and ultimately improves light energy utilization efficiency.

[0027] Synergistic regulation in response to "high temperature stress": When the diagnostic model outputs a state of "high temperature stress" (usually indicated by a sustained high canopy temperature and a decrease in the Fv / Fm value), and its confidence level is higher than a preset threshold, this strategy is triggered. The system will generate a synchronously executed collaborative control instruction package, which includes optical control instructions and microclimate control instructions. The optical control instructions use an attenuation coefficient β (the value of β can be set from 0.1 to 0.4) to synchronously set the PWM duty cycle of the red LED channel (600-700nm) and the far-red LED channel to (1 - β) times the current value, based on the degree of stress. The microclimate control instructions set the drive voltage of all fans integrated in the supplementary lighting fixture housing to the maximum value V_max, or to a high value calculated based on the temperature exceedance and higher than the base speed, so that the fans enter the highest speed or powerful operation state. Therefore, reducing the proportion of red and far-red light directly reduces the radiant energy absorbed by the leaves and converted into heat, thus reducing the canopy's radiative heat load at its source. Simultaneously, increasing fan speed enhances canopy airflow, significantly promoting leaf transpiration. The latent heat of water evaporation directly and rapidly lowers leaf surface temperature and effectively disrupts the static air boundary layer on the leaf surface, greatly reducing resistance to CO2 diffusion into the stomata. This ensures a sufficient supply of "raw materials" for the dark reaction of photosynthesis at high temperatures, preventing carbon starvation. Thus, "optical load reduction" (reducing heat absorption) and "ventilation enhancement" (strengthening heat dissipation and ensuring air supply) occur simultaneously in time and space, forming a powerful synergy against high-temperature stress. This coordinated intervention can rapidly reverse the physiological stress state of plants.

[0028] In one embodiment, the fan of the present invention can also serve as a multi-dimensional airflow controllable microclimate regulation module. Specifically, an electrically adjustable guide vane group is designed at the fan outlet. By controlling the deflection angle of these vanes through a program, the airflow can be precisely guided to different areas of the canopy (such as the top, middle, or bottom), rather than being blown uniformly across the entire canopy. At the same time, the fan's drive circuit is upgraded to support high-frequency pulse and frequency conversion control, ensuring that the fan can not only adjust the average speed but also deliver "pulse-type" airflow at a specific frequency and duty cycle (e.g., delivering airflow for 2 seconds, stopping for 1 second, and repeating this cycle). Furthermore, a ring of miniature ultrasonic atomizing nozzles is integrated around the fan inside the lamp housing. These nozzles can be controlled by the system to spray extremely fine water mist when needed. Therefore, when the AI ​​diagnoses a plant as being in a state of "decreased photosynthetic efficiency" and triggers the "increase far-red light" command, the system simultaneously activates the fan's "low-frequency pulse" mode. This promotes leaf movement through periodic gentle breezes, resulting in a more uniform light distribution within the canopy and reducing light spots and shadow effects. This, combined with the effect of far-red light promoting photosystem balance, more thoroughly optimizes the light energy distribution and utilization efficiency within the canopy. Its improvement effect is far superior to supplemental lighting alone. Compared to continuous airflow, pulsed airflow can also effectively prevent adaptive growth (such as mechanical damage or excessive transpiration) caused by plants being exposed to constant wind for extended periods. When "high" is diagnosed... During heat stress, the system not only executes commands to reduce the proportion of red / far-red light and the maximum rotation speed, but also controls the guide vanes to concentrate the airflow on the densest and hottest "hot spots" in the canopy. This targeted and precise cooling avoids wasting airflow and provides rapid relief to the core stress area with minimal energy consumption. At the same time, if an atomization system is integrated, a brief (e.g., 10-15 seconds) micro-spray can be initiated simultaneously at the beginning of the fan's full-speed operation. The water mist is dispersed by the strong airflow and blown towards the canopy, instantly reducing the leaf surface temperature through evaporative heat absorption, creating a "buffer period" for the plant to recover quickly from heat stress.

[0029] A dynamic spectral supplemental lighting device based on plant physiological signal feedback, comprising: A sensing unit is configured for non-contact acquisition of physiological optical signals from light-receiving plants. The sensing unit includes at least one of a multispectral imaging camera and a chlorophyll fluorometer. The sensing unit can be a composite probe integrating a multispectral camera and a chlorophyll fluorometer, or it can be two independent devices that communicate with the processing and control unit via wired (e.g., Ethernet) or wireless (e.g., Wi-Fi). To ensure diagnostic accuracy, the system software must handle the time synchronization of data acquisition from different sensors, ensuring that PRI data and Fv / Fm data used for diagnosis at the same time are acquired within the shortest possible time window.

[0030] The processing and control unit, communicatively connected to the sensing unit, receives the physiological optical signals, runs the plant physiological state diagnostic model, and generates spectral modulation commands. During system operation, the processing and control unit collects time-series data of the physiological optical signals and their corresponding spectral modulation commands, forming an incremental dataset. This incremental dataset is then used periodically to fine-tune the plant physiological state diagnostic model. The method used for fine-tuning the plant physiological state diagnostic model is online transfer learning.

[0031] Specifically, the fine-tuning process for online transfer learning is as follows: The system initially loads a pre-trained general diagnostic model on a large, diverse dataset. This model can be viewed as a combination of a feature extractor and a classifier. Before fine-tuning, the system modifies the model configuration file, freezing the front-end convolutional and pooling layers for convolutional neural networks and freezing the first few LSTM units for long short-term memory networks. These frozen layers are responsible for extracting shallow and deep general features (such as leaf texture, color distribution, and temporal variation patterns), and their parameters remain unchanged during fine-tuning to preserve the model's existing extensive plant physiological knowledge. The system only unlocks the last one or two fully connected layers (or classification layers) of the model. These layers are responsible for mapping the general features extracted from the front end to specific physiological state classifications (such as "healthy" and "light stress"). Subsequently, the system divides the locally collected incremental dataset into training and validation sets (e.g., at an 80% / 20% ratio) and retrains only the unlocked fully connected layer parameters with a significantly reduced learning rate (e.g., 1 / 10 to 1 / 100 of the initial training learning rate). The computational cost of this operation is far less than training a full model, making incremental learning possible on embedded platforms (such as NVIDIA Jetson series) or edge servers. Essentially, it allows the model to quickly learn how to map these fused features to the physiological states of specific local crops and environments while retaining its general visual / temporal pattern recognition capabilities. This ensures that the invention maintains optimal regulatory performance over the long term in any deployment environment. This mechanism ensures that the diagnostic model can continuously adapt to the local environment, thereby maintaining the accuracy and robustness of the system's long-term operation.

[0032] The supplementary lighting and microclimate coordinated control unit is communicatively connected to the processing and control unit, and includes multiple independently controllable spectral channels and a microclimate adjustment mechanism for responding to the spectral control commands; the supplementary lighting and microclimate coordinated control unit includes: a multispectral LED module, which includes at least deep red, far red and blue LED chip arrays; one or more fans, integrated with the LED module in the same housing, whose speed is independently controlled by the processing and control unit.

[0033] The supplementary lighting and microclimate control unit is an integrated intelligent lighting fixture. The LED chip array and fan are housed together in a housing with good heat dissipation and protection rating (such as IP65). Its internal circuit board contains multiple independent constant current drive circuits corresponding to different wavelengths of the LED array, as well as the fan motor drive circuit. All these circuits are under the unified command of the processing and control unit. This integrated design simplifies installation and wiring, ensures that the effective range of light and wind is consistent, and is the physical basis for achieving precise and coordinated control.

[0034] In summary, this invention achieves a fundamental shift from traditional "environment-driven" to "plant physiology-driven" approaches by directly collecting and analyzing physiological optical signals characterizing plant photosynthetic efficiency and health status, enabling supplemental lighting decisions to precisely match the plant's real-time needs. Through hardware integration and command synchronization of spectral modulation and canopy ventilation, a "light-air synergy" effect is created, simultaneously achieving heat load reduction, strong heat dissipation, and air supply preservation under high-temperature stress, overcoming the bottleneck of single-control mechanisms. Online transfer learning enables the system to possess personalized evolutionary capabilities, allowing it to continuously adapt to specific crops and environments to maintain long-term optimal performance. Finally, based on the deep physiological and optical mechanisms of plants, it achieves early warning of adversity and precise functional regulation starting from the energy balance of the photosystem, constructing an adaptive and synergistically efficient intelligent supplemental lighting system.

[0035] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. A dynamic spectrum light supplementing method based on plant physiological signal feedback, characterized in that, The method comprises the following steps: S1: collecting at least one physiological optical signal of the light-receiving plant by a non-contact sensing unit; S2: inputting the physiological optical signal into a preset plant physiological state diagnosis model to output a current physiological state judgment result of the light-receiving plant; S3: based on the current physiological state judgment result, generating a light spectrum regulation instruction containing light quality matching parameters according to a predefined light spectrum regulation strategy; S4: driving the independently light-adjustable multi-spectrum LED unit according to the light spectrum regulation instruction, and cooperatively regulating the plant canopy microclimate to dynamically supplement light for the light-receiving plant.

2. The dynamic spectral light supplementing method based on plant physiological signal feedback according to claim 1, characterized in that, In step S1, the physiological optical signal includes a photochemical reflectance index obtained by a multi-spectrum or hyperspectral imager, and / or a PSII maximum photochemical quantum yield obtained by a chlorophyll fluorometer. 3.The dynamic spectral light supplementing device and method based on plant physiological signal feedback according to claim 1, wherein, In step S2, the plant physiological state diagnosis model is a trained machine learning model.

4. The dynamic spectral light supplementing method based on plant physiological signal feedback according to claim 1, characterized in that, In step S4, the cooperative regulation of the plant canopy microclimate is realized by adjusting the rotating speed of a fan integrated with the multi-spectrum LED unit.

5. The dynamic spectral light supplementing method based on plant physiological signal feedback according to claim 1, characterized in that, In step S3, the predefined light spectrum regulation strategy includes: when the diagnosis model judges that the plant is in a photosynthetic efficiency decline state, generating an instruction to increase the proportion of far-red light radiation in the light spectrum; when the diagnosis model judges that the plant is in a high-temperature stress state, generating an instruction to simultaneously reduce the proportion of red light and far-red light and increase the rotating speed of the fan.

6. A dynamic spectral light supplementing device for implementing the method of any one of claims 1 to 5, characterized in that, It comprises: a sensing unit configured to non-contact collect physiological optical signals of a light-receiving plant; a processing and control unit communicatively connected to the sensing unit, configured to receive the physiological optical signals, run a plant physiological state diagnosis model, and generate a light spectrum regulation instruction; a light supplementing and microclimate cooperative regulation unit communicatively connected to the processing and control unit, comprising a plurality of independently controllable spectral channels and a microclimate adjusting mechanism, configured to respond to the light spectrum regulation instruction.

7. The dynamic spectral light supplement device based on plant physiological signal feedback according to claim 6, characterized in that, The sensing unit comprises at least one of a multi-spectrum imaging camera and a chlorophyll fluorometer. 8.The dynamic spectral light supplementing device based on plant physiological signal feedback according to claim 6, wherein, The light supplementing and microclimate cooperative regulation unit comprises: a multi-spectrum LED module comprising at least deep red, far red and blue LED chip arrays; one or more fans integrated with the LED module in the same housing, whose rotating speed is independently controlled by the processing and control unit.

9. The dynamic spectral light supplement device based on plant physiological signal feedback according to claim 6, characterized in that, The processing and control unit collects physiological optical signal time series data and corresponding light spectrum regulation instructions during system operation to form an incremental data set, and periodically fine-tunes the plant physiological state diagnosis model using the incremental data set.

10. The dynamic spectral light supplement device based on plant physiological signal feedback according to claim 9, characterized in that, The method for fine-tuning the plant physiological state diagnosis model is online transfer learning.

Images