An artificial intelligence-based plant light supplementing adaptive control method

Abstract

The application discloses a plant light supplementing self-adaptive control method based on artificial intelligence and belongs to the technical field of plant light regulation. The method first acquires input data related to the light supplementing demand of a target plant, synchronously inputs the data into a pre-trained AI light supplementing regulation model, calculates and outputs full-dimension light supplementing control parameters matched with the current growth demand of the target plant, generates control instructions according to the parameters, and adjusts the operation state of a plant light supplementing lamp to perform light supplementing operation. The application does not need manual adjustment of professional parameters, has no rigid limitation of hardware and data types, improves light supplementing efficiency and scene adaptability, reduces the use threshold, and can be widely applied to various plant light supplementing scenes.

Description

Technical Field

[0001] This invention belongs to the field of plant facility cultivation and intelligent light control technology, specifically relating to an artificial intelligence-based adaptive control method for plant supplemental lighting. IPC classification numbers: A01G7 / 04; H05B47 / 105; G05B13 / 04G06N3 / 04 Background Technology

[0002] With the rapid popularization of home gardening, indoor potted plant cultivation, and small-scale facility agriculture, insufficient natural light has become a core factor restricting the healthy growth of indoor plants. According to publicly available industry data, the number of indoor gardening users in my country has exceeded 30 million, with over 80% of these users experiencing insufficient light on their balconies or indoor spaces. Plant supplemental lighting has become a core necessity for bridging light gaps and regulating plant growth rhythms. Simultaneously, with the development of LED lighting and artificial intelligence technologies, optimizing plant supplemental lighting through intelligent means has become a core research and development direction in this field.

[0003] In existing technologies, plant grow light control schemes are mainly divided into two categories. The core content and objective technical defects of each scheme are as follows:

[0004] The first type is the manual preset fixed parameter control scheme. This is the mainstream control mode for commercially available residential supplemental lighting. Typical schemes can be found in the instruction manuals of commercially available supplemental lighting products. Its core technology involves the user manually setting the on / off time, fixed light intensity, and light quality ratio of the supplemental lighting in advance. The supplemental lighting can only perform simple timed and quantitative supplemental lighting according to the preset parameters. The core technical drawbacks of this scheme are: firstly, it cannot dynamically adjust the supplemental lighting strategy according to the differences in plant varieties and growth stages, and cannot match the differentiated light requirements of different stages such as germination, vegetative growth, flowering, and fruiting. This easily leads to problems such as insufficient supplemental lighting causing excessive vegetative growth, or excessive supplemental lighting causing light damage and leaf burn. The supplemental lighting effectiveness is less than 40%. Secondly, it requires users to have professional knowledge of plant photochemistry and plant physiology to set reasonable supplemental lighting parameters, making it extremely difficult to use and completely unsuitable for ordinary individual users without a professional background.

[0005] The second category is intelligent supplemental lighting control solutions. These solutions incorporate sensors, automated control, and even artificial intelligence technologies to optimize the supplemental lighting effect. Representative patented solutions currently disclosed are as follows:

[0006] Chinese invention patent application CN116491324A discloses a low-carbon, energy-saving supplemental lighting method based on plant light requirements. Its core scheme involves collecting light data from the plant's environment, inputting this data into a supplemental lighting decision model to output a target light value, and then controlling the operation of the supplemental lighting lamps. However, this scheme uses only environmental light data as the sole input to the model, failing to incorporate plant growth characteristics such as variety and growth cycle. Therefore, it cannot achieve differentiated supplemental lighting adaptation for different varieties and growth stages, and the accuracy of supplemental lighting remains significantly limited.

[0007] Chinese invention patent CN119946951B discloses an AI-based plant light spectral adjustment method for optimizing plant photosynthesis. Its core scheme involves collecting plant physiological data to construct a three-dimensional spectral response model and training a network through knowledge distillation to generate an LED dimming scheme. However, this scheme only focuses on adjusting the single dimension of the supplemental lighting spectrum, failing to cover the coordinated control of all supplemental lighting parameters such as switching timing, light intensity, and illumination angle. Furthermore, it relies on customized plant physiological data acquisition equipment and is not compatible with commercially available supplemental lighting devices used by ordinary individual users.

[0008] Chinese invention patent application CN112867196A discloses a method for implementing a plant light formula supplemental lighting system based on artificial intelligence. Its core solution involves determining the plant species and growth stage through image recognition, retrieving fixed light formula data from a preset database, and adjusting the output of the supplemental lighting. This solution is essentially a fixed formula model of "image recognition + database lookup matching," lacking the autonomous prediction and adaptive control capabilities of an AI model. It cannot adapt to dynamically changing environmental conditions and plant growth states, and thus remains limited by the adaptability constraints of a fixed formula.

[0009] Chinese invention patent CN113994829B discloses a method for controlling the operation of LED supplemental lighting that considers time-shifting and cost factors. Its core solution involves predicting the next day's illumination curve using a BP neural network to determine the optimal supplemental lighting time and power. While this solution aims to optimize energy consumption and cost, it only adjusts two parameters: supplemental lighting time and power. It fails to achieve precise control of all supplemental lighting parameters tailored to plant growth needs, thus failing to address the core problem of the mismatch between supplemental lighting effect and plant growth requirements.

[0010] Through a comprehensive analysis of the aforementioned existing technologies, the closest existing technology in this field is the technical solution published under CN116491324A. Although this solution achieves intelligent control of supplemental lighting based on environmental data, it still suffers from common technical defects that cannot be overcome: First, it imposes rigid and fixed limitations on the data types of the input model, supporting only a few types of data explicitly listed in the solution, and cannot be compatible with other multi-dimensional data collection data related to plant supplemental lighting needs that already exist in this field, thus the solution has inherent limitations in scenario adaptability and versatility; Second, the control dimension is singular, failing to achieve coordinated adaptive control of all-dimensional supplemental lighting parameters, resulting in insufficient matching between the supplemental lighting effect and plant growth needs; Third, the AI ​​model is merely a simple scenario transfer of a general model, without targeted adaptation design for plant supplemental lighting scenarios, and cannot solve the core problem of real-time matching between the dynamic light requirements of plants and supplemental lighting parameters; Fourth, it has a high barrier to entry, relying on customized hardware or professional knowledge, and cannot be adapted to ordinary individual users without professional backgrounds. Summary of the Invention

[0011] I. Technical problems to be solved

[0012] In view of the objective defects of the existing technical solutions cited in the background art, the present invention aims to solve the following core technical problems and achieve the corresponding technical objectives by deeply integrating artificial intelligence technology with plant supplemental lighting technology:

[0013] Breaking through the rigid and fixed limitations of existing technologies on the data types of input models, it is compatible with all existing multi-dimensional data collection related to plant supplemental lighting needs, solving the problems of poor scenario adaptability and insufficient versatility of existing technical solutions;

[0014] Breaking through the limitations of existing technologies that rely on a single dimension for supplemental lighting control, this technology enables coordinated and adaptive control of all dimensions of supplemental lighting parameters, including on / off timing, light intensity, light quality ratio, illumination angle, and total effective illumination duration per day. This fundamentally avoids the problems of insufficient or excessive supplemental lighting and improves the matching degree between supplemental lighting accuracy and plant growth needs.

[0015] Complete the targeted adaptation design of artificial intelligence models for plant supplemental lighting scenarios, and solve the core technical problem that the dynamic light requirements of plants and supplemental lighting parameters cannot be matched in real time due to the simple transfer of general AI models in existing technologies;

[0016] Completely eliminate the barriers to using supplemental lighting solutions, solve the problem of existing technologies relying on customized hardware and professional plant lighting knowledge, and enable ordinary individual users without professional background or engineering implementation capabilities to easily use optimal supplemental lighting control;

[0017] It enables autonomous closed-loop iterative optimization of supplemental lighting schemes, solving the problem that existing technologies lack autonomous learning and optimization capabilities. This allows the supplemental lighting strategy to be continuously upgraded based on the actual growth effect of plants, maintaining the optimal supplemental lighting effect without human intervention.

[0018] It adapts to scenarios where multiple plant varieties are planted in the same area, solves the problem that existing technologies cannot simultaneously adapt to the light requirements of different plant varieties, realizes differentiated zoned supplemental lighting for multiple plant varieties, and broadens the applicable scenarios of the technical solution.

[0019] This enriches the implementation methods of AI-based supplemental lighting control models, addresses the issues of high barriers to entry and insufficient flexibility in deployment of existing technology models, and adapts to the hardware conditions and usage needs of different users.

[0020] II. Technical Solution

[0021] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0022] (I) Overall Technical Solution

[0023] An artificial intelligence-based adaptive control method for plant supplemental lighting, applied to a supplemental lighting control system comprising plant supplemental lighting, a data acquisition unit, an AI computing unit, and a main control unit, includes the following steps:

[0024] S1. Obtain input data related to the supplemental lighting requirements of the target plant, including growth-related data of the target plant and environmental monitoring data of the growth environment of the target plant;

[0025] S2. The input data related to the supplemental lighting needs of the target plant are synchronously input into the pre-trained AI supplemental lighting control model, and the AI ​​supplemental lighting control model is used to calculate and output supplemental lighting control parameters that match the current growth needs of the target plant;

[0026] S3. Generate control commands based on the supplemental lighting control parameters, and send them to the plant supplemental lights through the main control unit to adjust the operating status of the plant supplemental lights to perform supplemental lighting operations.

[0027] The technical boundaries of the input data related to the supplemental lighting needs of the target plant are: all collectible data that directly affects the supplemental lighting effect of the plant and is related to plant photosynthesis and growth and development, including but not limited to plant growth-related data and environmental monitoring data listed in this solution.

[0028] (ii) Additional technical solutions

[0029] 1. In step S1, the growth-related data of the target plant includes at least one of the following: plant variety, current growth cycle stage, and plant photophysiological parameters.

[0030] 2. In step S1, the environmental monitoring data includes at least one of the following: ambient natural light intensity, natural light quality composition, ambient light duration, ambient temperature, ambient humidity, and ambient carbon dioxide concentration.

[0031] 3. In step S2, the supplemental lighting control parameters include at least one of the following: the on / off sequence of the plant supplemental lighting, light intensity, light quality ratio, irradiation angle, and total effective irradiation time per day.

[0032] 4. In step S2, the pre-trained AI supplemental lighting control model is a regression prediction model trained by a deep learning neural network using standard light requirements parameters and corresponding growth effect data of different plant varieties under different growth cycles and environmental conditions as the training set.

[0033] 5. The AI ​​supplemental lighting control model adopts a conventional fully connected neural network architecture in the art. The input layer is a normalized feature vector of the input data related to the supplemental lighting requirements of the target plant. The hidden layer uses a conventional 3-5 fully connected layer to extract features. The output layer is the prediction result of the supplemental lighting control parameters. The model is trained using the conventional gradient descent method and backpropagation algorithm in the art. The training set uses standard data from publicly available plant lighting databases known in the art, including but not limited to the plant lighting database of the Chinese Academy of Agricultural Sciences and the plant light requirements database published by the International Society for Horticultural Science. Those skilled in the art can complete the model construction and training without creative effort.

[0034] 6. This method also includes a closed-loop optimization step S4: collecting growth feedback data related to the supplemental lighting effect of the target plant after the supplemental lighting operation is performed, inputting the growth feedback data into the AI ​​supplemental lighting control model, iteratively updating the model, and optimizing the output strategy of subsequent supplemental lighting control parameters.

[0035] 7. The growth feedback data related to the supplemental lighting effect of the target plant includes at least one quantifiable growth data among plant height, leaf area, leaf color, flowering status, and fruiting status acquired through the image acquisition unit.

[0036] 8. In step S4, the specific method of model iterative update is as follows: the collected growth feedback data is compared with the preset plant growth target threshold, the network weight parameters of the AI ​​supplemental lighting control model are adjusted according to the comparison results, and the output logic of the supplemental lighting control parameters is corrected.

[0037] 9. When there are multiple target plants of different varieties in the supplemental lighting area, the input data related to the supplemental lighting requirements of each target plant is obtained. The AI ​​supplemental lighting control model outputs differentiated supplemental lighting control parameters that match each target plant, and controls the plant supplemental lights in the corresponding zones to perform independent supplemental lighting operations.

[0038] 10. The AI ​​supplementary lighting control model is deployed on a cloud computing platform, and the main control unit completes data upload and acquisition of supplementary lighting control parameters through a wireless network; or the AI ​​supplementary lighting control model is deployed on a local embedded computing module, and completes data calculation and parameter output locally.

[0039] 11. The AI ​​supplemental lighting control model can be obtained by using a commercially available open-source basic model and then fine-tuning it using a dedicated dataset for the field of plant supplemental lighting. The dedicated dataset includes standard light requirements parameters and corresponding growth effect data for different plant varieties under different growth cycles and environmental conditions. Those skilled in the art can complete the fine-tuning and adaptation of the model without creative effort.

[0040] III. Beneficial Effects

[0041] Compared with the prior art cited in the background section, this invention has prominent substantive features and significant progress. All beneficial effects are directly brought about and inevitably generated by the corresponding technical features of the above-mentioned technical solution, and correspond one-to-one with the technical problem to be solved, as detailed below:

[0042] This invention breaks through the fixed data type limitations of existing technologies, significantly improving the scenario adaptability and versatility of the solution. Using input data related to the supplemental lighting needs of the target plant as the model's input basis, this invention overcomes the rigid limitations of existing technologies on input data types. It is compatible with all existing plant growth data and environmental monitoring data related to supplemental lighting, fundamentally solving the problems of insufficient adaptability and poor versatility caused by fixed data types in prior art such as CN116491324A. This significantly improves the technical solution's adaptability to different planting scenarios and plant varieties.

[0043] 1. This invention achieves coordinated and adaptive regulation of all-dimensional supplemental lighting control parameters, fundamentally solving the core technical problems of insufficient or excessive supplemental lighting, and significantly improving the accuracy and effectiveness of supplemental lighting. By simultaneously acquiring all-dimensional input data related to plant supplemental lighting needs and combining it with an AI supplemental lighting regulation model designed specifically for plant needs, this invention directly outputs all-dimensional supplemental lighting control parameters that precisely match the plant's current growth requirements. This overcomes the limitations of existing technologies that can only achieve single threshold triggering, single parameter adjustment, and fixed formula matching. It can dynamically adjust the supplemental lighting strategy according to differences in plant variety and growth cycle, perfectly matching the differentiated light requirements of plants at different growth stages, and completely avoiding the problems of excessive vegetative growth caused by insufficient supplemental lighting and light damage caused by excessive supplemental lighting. Based on publicly available LED supplemental lighting experimental data from the Institute of Environmental Development, Chinese Academy of Agricultural Sciences, this dynamic supplemental lighting strategy improves the supplemental lighting effectiveness by more than 60% and the energy utilization efficiency by 2.6 times compared to existing fixed-parameter supplemental lighting schemes.

[0044] 2. This invention achieves deep integration of artificial intelligence and plant supplemental lighting technology, completing a targeted adaptation design for plant supplemental lighting scenarios. Addressing the specific technical needs of multi-dimensional parameter control in plant supplemental lighting, this invention designs an AI supplemental lighting control model based on multi-dimensional supplemental lighting-related input data and outputting supplemental lighting control parameters. This is not a simple scenario transfer of general AI models in existing technologies such as CN112867196A. From model input / output design, training logic, and application scenarios, this invention completes a dedicated adaptation for plant supplemental lighting needs, directly solving the core technical problem of real-time matching of dynamic light requirements of plants with supplemental lighting parameters.

[0045] 3. Completely eliminates the barrier to entry for this solution, making it fully compatible with ordinary individual users without professional background or engineering implementation capabilities. This invention automatically generates supplemental lighting control parameters through an AI model, eliminating the need for users to have professional knowledge of plant photology, repeatedly adjust supplemental lighting parameters manually, or customize hardware. It is directly compatible with most mainstream commercially available plant grow lights, solving the core defects of existing technologies that have high barriers to entry, rely on professional knowledge, and require customized hardware. Ordinary individual users can achieve optimal supplemental lighting control for plants without any professional background or engineering implementation capabilities.

[0046] 4. Significantly reduces the energy consumption of the supplemental lighting system, demonstrating outstanding energy-saving and environmental protection effects. This invention dynamically adjusts the supplemental lighting strategy through an AI model combined with real-time input data from all dimensions. It can adjust the operating status of the supplemental lights in real time according to changes in natural light intensity and light quality, outputting matching supplemental lighting control parameters only during the periods when plants need them, significantly reducing the duration of ineffective supplemental lighting and redundant energy consumption. Based on publicly available greenhouse supplemental lighting experimental data from Wageningen University in the Netherlands, similar dynamic adaptive supplemental lighting schemes can significantly reduce the energy consumption of the supplemental lighting system compared to fixed-parameter supplemental lighting schemes, while simultaneously increasing crop yields, combining economic efficiency and environmental friendliness.

[0047] 5. Achieve autonomous closed-loop optimization of the supplemental lighting scheme, with the supplemental lighting effect continuously improving as the plant grows. This invention collects plant growth feedback data related to the supplemental lighting effect and iteratively updates the AI ​​supplemental lighting control model, enabling the supplemental lighting strategy to continuously and autonomously optimize according to the actual growth effect of the plant. It can always maintain the optimal supplemental lighting effect without human intervention, thus solving the deficiency of existing technologies that lack autonomous learning and optimization capabilities.

[0048] 6. Supports simultaneous and precise supplemental lighting for multiple plant varieties, further expanding applicable scenarios. This invention addresses scenarios where multiple plant varieties are planted in the same area. Based on the supplemental lighting-related input data for each plant, it outputs differentiated supplemental lighting control parameters to achieve independent supplemental lighting for each zone. This solves the problem that existing technologies cannot simultaneously adapt to the light requirements of different plant varieties. It is widely applicable to various supplemental lighting scenarios such as home balcony gardening, indoor potted plant cultivation, small plant factories, and greenhouses, demonstrating extremely high versatility.

[0049] 8. The model offers flexible and diverse implementation methods, significantly expanding the applicable scenarios and ease of implementation. The AI ​​supplemental lighting control model of this invention can be built from scratch using conventional fully connected neural networks, or it can be obtained by fine-tuning a widely available open-source basic model. This adapts to the needs of different stakeholders, including individual users, small-scale growers, and large-scale greenhouses. It eliminates the need for developing a model architecture from scratch; scenario adaptation can be quickly achieved based on a general-purpose model, significantly lowering the threshold for implementation. It is widely compatible with current mainstream AI computing environments, significantly improving its industrial applicability and versatility. Attached Figure Description

[0050] Figure 1 is a flowchart of an artificial intelligence-based adaptive control method for plant supplemental lighting according to the present invention. Detailed Implementation

[0051] The present invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art will understand that sensors, main control units, communication protocols, AI model training and fine-tuning methods, etc., not described in detail in this embodiment, can all be implemented using conventional commercially available products and conventional technical means in the field, without requiring creative effort.

[0052] This invention provides an artificial intelligence-based adaptive control method for plant supplemental lighting, applied to a supplemental lighting control system comprising a plant supplemental light, a data acquisition unit, an AI computing unit, and a main control unit. The data acquisition unit can collect various input data related to the supplemental lighting needs of the target plant, including commercially available environmental sensors, image acquisition devices, and plant physiological monitoring devices. The AI ​​computing unit can utilize a general-purpose cloud computing platform or a commercially available local embedded computing module, requiring no specific hardware configuration. The main control unit employs a conventional microcontroller, microprocessor, or smart gateway device, capable of wireless or wired communication to connect with other units and is compatible with the driving control protocols of most mainstream plant supplemental lights on the market.

[0053] Example 1: Basic Implementation

[0054] This embodiment is a basic implementation of the present invention, fully covering the protection scope of the independent claims, and specifically includes the following steps:

[0055] S1. Data acquisition steps: The main control unit acquires input data related to the supplemental lighting requirements of the target plant. The input data includes growth-related data of the target plant and environmental monitoring data of the growth environment in which the target plant is located.

[0056] Among them, the growth-related data of the target plant is input by the user through conventional interactive terminals such as mobile APP and computer web page, including plant variety (such as rose, succulent, tomato, etc.), current growth cycle stage (such as germination period, vegetative growth period, flowering period, fruiting period), and photophysiological parameters of the plant's preference for shade / sun; environmental monitoring data is collected in real time through corresponding conventional commercially available sensors, including ambient natural light intensity, ambient temperature, and ambient humidity.

[0057] S2. AI Decision-Making Steps: The main control unit synchronously inputs the acquired input data related to the supplemental lighting needs of the target plant into the pre-trained AI supplemental lighting control model, and outputs supplemental lighting control parameters that match the current growth needs of the target plant through model calculation.

[0058] The AI ​​supplemental lighting control model is a pre-trained deep learning neural network regression prediction model. The training set uses standard data from a publicly available plant photodynamic database, including the optimal light requirements and corresponding growth effects of different plant varieties under different growth cycles and environmental conditions. The model is trained through supervised learning. The input layer is a normalized feature vector of the input data related to the supplemental lighting requirements of the target plant. The hidden layer uses four fully connected layers to extract features. The output layer is the prediction result of the supplemental lighting control parameters. In this embodiment, the supplemental lighting control parameters include the on / off sequence of the plant supplemental lighting lamp, light intensity, red-blue light quality ratio, and total effective irradiation time per day.

[0059] S3. Supplemental Lighting Execution Steps: The main control unit generates corresponding control commands based on the supplemental lighting control parameters output by the AI ​​supplemental lighting control model, and sends them to the driver module of the plant supplemental light via WiFi wireless communication to adjust the on / off status, light intensity, and light quality ratio of the supplemental light, and execute the corresponding supplemental lighting operation.

[0060] Example 2: Preferred Implementation with Closed-Loop Optimization

[0061] This embodiment is a preferred implementation based on Embodiment 1, covering additional technical solutions related to closed-loop optimization. After step S3 in Embodiment 1, it also includes the following steps:

[0062] S4. Closed-loop optimization steps, specifically divided into:

[0063] S4-1. Growth Data Acquisition: Through conventional image acquisition units such as cameras, periodically collect growth feedback data related to the supplemental lighting effect on the target plant after the supplemental lighting operation is performed, including quantifiable growth data such as plant height, leaf area, leaf color, and flowering status;

[0064] S4-2. Model Iteration and Update: The collected growth feedback data is compared with the preset plant growth target threshold to determine the actual effect of the current supplemental lighting strategy. Based on the comparison results, the network weight parameters of the AI ​​supplemental lighting regulation model are adjusted through the backpropagation algorithm commonly used in this field, and the output logic of the supplemental lighting control parameters is corrected to realize the autonomous iterative optimization of the model. This ensures that the subsequent output supplemental lighting control parameters continuously meet the actual growth needs of the target plant, forming a complete closed-loop regulation system.

[0065] Example 3: Extended Implementation of Multi-Plant Zone Supplemental Lighting

[0066] This embodiment is an extended implementation, covering an additional technical solution for zoned supplemental lighting of multiple plant varieties. When multiple target plants of different varieties exist in the supplemental lighting area, the area is divided into independent zones corresponding to each plant. Each zone is equipped with an independent data acquisition unit and a plant supplemental light. Input data related to the supplemental lighting requirements of each target plant is acquired. The AI ​​supplemental lighting control model outputs differentiated supplemental lighting control parameters that match the growth requirements of each target plant. The main control unit generates control commands based on the parameters of each zone, controlling the plant supplemental lights in the corresponding zone to perform independent supplemental lighting operations, thereby achieving synchronous and precise supplemental lighting for multiple plant varieties.

[0067] Example 4: Optional Implementation Methods for Model Deployment

[0068] This embodiment is an optional implementation of the model deployment, covering additional technical solutions related to model deployment. The AI ​​supplemental lighting control model is deployed on a general cloud AI computing platform. The main control unit uploads the input data related to the supplemental lighting needs of the target plant to the cloud computing platform via wireless networks such as 4G / 5G or WiFi. After the cloud platform completes the model calculation, it sends the supplemental lighting control parameters back to the main control unit. No dedicated AI computing hardware needs to be configured locally, further reducing the usage threshold and hardware costs for individual users.

[0069] Example 5: Preferred Implementation Based on Fine-tuning of a General Basic Large Model

[0070] This embodiment is a preferred implementation of the AI ​​supplementary lighting control model, covering additional technical solutions related to fine-tuning of the general basic model, as detailed below:

[0071] The AI ​​supplemental lighting control model adopts a commercially available open-source basic model, such as Llama 3, DeepSeekV3, or Tongyi Qianwen open-source version, which are well-known and freely available in the field. It is trained by targeted fine-tuning using a dataset specifically designed for plant supplemental lighting.

[0072] The dedicated dataset for fine-tuning training uses publicly available standard data from plant lighting databases that are well-known in the field, including data from the Chinese Academy of Agricultural Sciences Plant Lighting Database and the International Society for Horticultural Science's publicly available plant light requirements database, as well as optimal light requirements parameters and corresponding growth effect data under different environmental conditions. The fine-tuning training adopts the supervised fine-tuning (SFT) method, which is conventional in the field, with the accurate prediction of plant supplemental lighting control parameters as the core fine-tuning goal. Those skilled in the art can complete the fine-tuning adaptation and deployment of the model without creative labor.

[0073] After the model is fine-tuned, input data related to the supplemental lighting needs of the target plant is input, and the supplemental lighting control parameters that match the current growth needs of the plant can be output, thus realizing the adaptive regulation of plant supplemental lighting.

[0074] Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions in the specification are only for the purpose of understanding the principles of the present invention. Various changes and modifications can be made to the present invention without departing from the spirit and scope of the present invention, and all such changes and modifications fall within the scope of protection claimed by the present invention.

Claims

1. An artificial intelligence-based adaptive control method for plant supplemental lighting, applied to a supplemental lighting control system comprising plant supplemental lighting, a data acquisition unit, an AI computing unit, and a main control unit, characterized in that, Includes the following steps: S1. Obtain input data related to the supplemental lighting requirements of the target plant, including growth-related data of the target plant and environmental monitoring data of the growth environment of the target plant; S2. The input data related to the supplemental lighting needs of the target plant are synchronously input into the pre-trained AI supplemental lighting control model, and the AI ​​supplemental lighting control model is used to calculate and output supplemental lighting control parameters that match the current growth needs of the target plant; S3. Generate control commands based on the supplemental lighting control parameters, and send them to the plant supplemental lights through the main control unit to adjust the operating status of the plant supplemental lights to perform supplemental lighting operations.

2. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, In step S1, the growth-related data of the target plant includes at least one of the following: plant variety, current growth cycle stage, and plant photophysiological parameters.

3. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, In step S1, the environmental monitoring data includes at least one of the following: ambient natural light intensity, natural light quality composition, ambient light duration, ambient temperature, ambient humidity, and ambient carbon dioxide concentration.

4. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, In step S2, the supplemental lighting control parameters include at least one of the following: the on / off sequence of the plant supplemental lighting, light intensity, light quality ratio, irradiation angle, and total effective irradiation time per day.

5. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, In step S2, the pre-trained AI supplemental lighting control model is a regression prediction model trained by a deep learning neural network using standard light requirements parameters and corresponding growth effect data of different plant varieties under different growth cycles and environmental conditions as the training set.

6. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 5, characterized in that, The input layer of the AI ​​supplemental lighting control model is a normalized feature vector of input data related to the supplemental lighting requirements of the target plant, and the output layer is the prediction result of the supplemental lighting control parameters.

7. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, The method further includes a closed-loop optimization step S4: collecting growth feedback data related to the supplemental lighting effect of the target plant after the supplemental lighting operation is performed, inputting the growth feedback data into the AI ​​supplemental lighting control model, iteratively updating the model, and optimizing the output strategy of subsequent supplemental lighting control parameters.

8. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 7, characterized in that, The growth feedback data related to the supplemental lighting effect on the target plant includes at least one quantifiable growth data among plant height, leaf area, leaf color, flowering status, and fruiting status, acquired through the image acquisition unit.

9. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 7, characterized in that, The specific method of the iterative update is as follows: the collected growth feedback data is compared with the preset plant growth target threshold, the network weight parameters of the AI ​​supplemental lighting control model are adjusted according to the comparison results, and the output logic of the supplemental lighting control parameters is corrected.

10. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, When there are multiple target plants of different varieties in the supplemental lighting area, the input data related to the supplemental lighting needs of each target plant is obtained. The AI ​​supplemental lighting control model outputs differentiated supplemental lighting control parameters that match each target plant, and controls the plant supplemental lights in the corresponding area to perform independent supplemental lighting operations.

11. The plant supplemental lighting adaptive control method based on artificial intelligence according to claim 1, characterized in that, The AI ​​supplemental lighting control model is deployed on a cloud computing platform, and the main control unit completes data upload and acquisition of supplemental lighting control parameters through a wireless network; or the AI ​​supplemental lighting control model is deployed on a local embedded computing module, and completes data calculation and parameter output locally.

Images