Plant factory illumination adjusting system and method based on growth period intelligent identification

Abstract

The invention relates to a plant factory illumination adjustment system and method based on growth period intelligent identification. The system comprises a background management module, a growth detection module, a data storage module and an intelligent regulation and control module which are connected in sequence. The background management module receives the real-time image and environment data of leaf vegetable growth and inputs the real-time image and environment data to the growth detection module; the growth detection module performs image preprocessing on the real-time image, intelligently identifies the growth period of the current leaf vegetable, predicts the growth information of the current leaf vegetable, and sends the growth information to the data storage module for storage and recording; the data storage module automatically matches an optimal light supplementing strategy suitable for the current growth information of the leaf vegetables according to a pre-configured staged light supplementing formula; and the intelligent regulation and control module dynamically regulates parameters of the execution mechanism based on the optimal light supplement strategy, and maintains the optimal illumination environment of the leaf vegetables in real time. Compared with the prior art, the method has the advantages of rapid and accurate regulation, effective energy conservation and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural production environment regulation, and in particular to a plant factory lighting regulation system and method based on intelligent identification of growth periods. Background Art

[0002] The light saturation point is the light intensity when the plant reaches the maximum net photosynthetic rate. Lighting that exceeds the light saturation point will cause light inhibition and energy loss. At the same time, in a closed multi-layer vertical cultivation system, it will also cause edge burning in the production of green leafy vegetables. Excessive light supplementation is not only an important factor in the high energy consumption of closed vertical agricultural production, but also affects the quality of green leafy vegetables. At present, most of the traditional light supplementation methods in plant factories are managed by constant light intensity and manual timing operation. Although the method is simple and effective, it ignores the differences in operating environment and crop needs. Affected by growth parameters such as temperature, carbon dioxide concentration, and humidity, there are large differences in the demand and absorption of light for crops in different growth periods. This difference will lead to different optimal light supplementation strategies for crops in different growth periods. Therefore, by identifying the optimization of the light supplementation strategy decision during the crop growth period, the ineffective light supplementation energy loss can be effectively reduced, which has the effect of energy saving in the greenhouse. If the changes in light intensity in the external environment can be considered, differentiated and intelligent control strategies can be formulated for different growth stages of crops, so that the light emission of LEDs can be quickly adjusted with changes in the environment, which can not only adapt to crop growth, but also further effectively save electricity.

[0003] At present, there are two main methods for identifying crop growth stages: manual observation and classical image processing. Manual observation has the disadvantages of strong subjectivity, low efficiency, and poor real-time performance, which is not conducive to real-time regulation and automated production of crop growth parameters in vertical agriculture. Classical image processing methods are usually based on two-dimensional image data taken during crop growth. After image segmentation, crop projection area, plant height, leaf size, leaf area, population coverage index and other indicators are extracted, and the mathematical relationship between growth parameters and growth stages is established through regression analysis and other methods. This method cannot well meet the needs of real-time non-destructive testing of crops in plant factories because the image segmentation and feature extraction algorithms are difficult to cope with complex scenes, have poor generalization ability, complex algorithm design, and usually cause a certain degree of damage to plants. In recent years, research has used computer vision technology based on deep learning for crop development period detection, which can effectively save labor, avoid manual detection that damages crops and human subjective errors, and can achieve continuous observation, and the real-time performance of observation can also be improved.

[0004] With the development of deep learning technology, many scholars at home and abroad have begun to use deep learning artificial intelligence network training models to solve various complex problems. The target recognition algorithm based on deep learning is applied in the field of crop growth period recognition. It is suitable for various types of crops and can quickly and accurately complete the recognition of crops at various growth stages. Commonly used target recognition algorithms are divided into deep models based on convolutional neural networks (such as ResNet, EfficientNet, VggNet, InceptionNet, Faster RCNN, SSD (Single Shot MultiBox Detector), YOLO series, etc.) and deep models based on Transformer (such as Swin-Transformer, DETR (Detection Transformer)). However, the traditional convolutional neural network method is greatly affected by interference factors such as the background conditions and lighting of the crop growth environment, especially in the recognition of small targets, complex background occlusion and other scenes. The accuracy and precision of recognition are low; although the visual Transformer method has excellent performance in processing tasks such as target detection of different scales and contextual semantic analysis, its model is relatively complex, the number of parameters required for training is large, it is not easy to converge, and the detection real-time performance is poor.

[0005] Therefore, by using image processing and deep learning technology, combined with traditional convolutional neural networks and the Transformer model in natural language processing models, the recognition of crop growth periods in vertical agricultural environments can be studied. This can reduce costs while achieving better prediction results, which is suitable for application needs in crop growth period recognition and prediction.

[0006] How to achieve rapid and intelligent adjustment of lighting in plant factories to enhance the adjustment effect has become a technical problem that needs to be solved. Summary of the invention

[0007] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a plant factory lighting adjustment system and method based on intelligent identification of growth periods.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] According to one aspect of the present invention, there is provided a plant factory lighting regulation system based on intelligent identification of growth periods, the system comprising a background management module, a growth detection module, a data storage module and an intelligent control module connected in sequence;

[0010] After the background management module receives the real-time RGB image, depth image and environmental data of the leafy vegetable growth, it inputs them into the growth detection module;

[0011] The growth detection module performs image preprocessing on the real-time image, intelligently identifies the growth period of the current leafy vegetable, predicts the growth information of the current leafy vegetable, and sends it to the data storage module for storage and recording;

[0012] The data storage module automatically matches the best lighting strategy suitable for the current growth information of leafy vegetables according to the pre-configured phased lighting formula;

[0013] The intelligent control module dynamically adjusts the parameters of the actuator based on the optimal supplementary lighting strategy to maintain the optimal lighting environment for leafy vegetables in real time.

[0014] Preferably, the growth detection module includes a growth period recognition model and a growth parameter prediction model;

[0015] The growth period recognition model is based on the RGB image of the leafy vegetable growth to identify the growth period of the leafy vegetable;

[0016] The growth parameter prediction model predicts the growth information of leafy vegetables based on the depth image and RGB image of the leafy vegetable growth.

[0017] More preferably, the growth period recognition model is an improved YOLO model, including a backbone network, a feature fusion network and a prediction head;

[0018] A C2F convolution module is introduced into the backbone network to extract image features. In the feature fusion network, the original PAFPN structure is replaced with a BiFPN structure to fuse feature information of different scales. A self-attention mechanism CTF module and a Transformer encoder structure are added to the prediction head to perceive the details of leafy vegetable images by calculating the query matrix between different channels.

[0019] More preferably, the growth parameter prediction model is an improved multimodal image fusion network, comprising an encoder module, a multimodal fusion module and a decoder module;

[0020] The encoder module includes a shared feature encoder based on a Restormer block, and a Transformer-based basic encoder and a CNN-based detail encoder connected to the shared feature encoder;

[0021] The multimodal fusion module is used to fuse the features output by the basic encoder and the detail encoder;

[0022] The decoder module uses the Restormer block as the basic unit of the decoder.

[0023] More preferably, the shared feature encoder extracts shared features from the high-resolution image by applying self-attention on the feature dimension, thereby extracting cross-modal shallow features;

[0024] The basic encoder uses a Transformer with spatial self-attention to extract long-range dependency features from the shared features output by the shared feature encoder;

[0025] The detail encoder extracts high-frequency detail information from the shared features output by the shared feature encoder based on CNN.

[0026] Preferably, the growth information of the leafy vegetables includes leaf area, stem height and dry and wet weight of the leafy vegetables.

[0027] Preferably, the actuator parameters include the voltage, red and blue light ratio and photoperiod of the LED lamp.

[0028] Preferably, the system further comprises a digital binocular camera and a sensor, wherein the digital binocular camera is used to collect real-time images of the growth of leafy vegetables; and the sensor is used to collect environmental data of the growth of leafy vegetables.

[0029] Preferably, the background management module monitors the growth status of leafy vegetables in real time, and intelligently adjusts the lighting strategy that meets the current growth requirements of leafy vegetables, as well as the commands of the control actuators.

[0030] According to another aspect of the present invention, a method for adjusting light intensity in a plant factory based on intelligent identification of growth periods is provided, the method comprising the following steps:

[0031] Step 1, collecting images and environmental data of leafy vegetable growth in real time through digital binocular cameras and sensors;

[0032] Step 2: The background management module monitors the growth status of leafy vegetables in real time and manages production remotely;

[0033] Step 3, the growth detection module intelligently identifies the growth period of leafy vegetables based on the image and predicts the growth information of leafy vegetables;

[0034] Step 4: The data storage module automatically matches the best lighting strategy suitable for the current growth information of leafy vegetables according to the pre-configured phased lighting formula;

[0035] Step 5: The intelligent control module controls the actuator to adjust the operating parameters based on the optimal fill light strategy.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1) The present invention utilizes a growth detection module to intelligently identify the current growth stage of leafy vegetables and predict the growth information of the current leafy vegetables, adds the factors of growth stage and growth information to the light supplement strategy of leafy vegetables, and automatically matches the light supplement formula according to the actual lighting demand required for the current growth stage of leafy vegetables. The output light supplement strategy is faster and more accurate than manual identification and regulation, and the environmental parameters are adjusted by the execution structure, which is beneficial to energy saving in plant factories and improves production management efficiency and the economic benefits of planting.

[0038] 2) The growth detection module of the present invention includes an improved YOLO model and an improved multimodal image fusion network, which reduces the amount of calculation while enhancing the model's detail feature extraction capability and key information conversion capability.

[0039] 3) The present invention combines leafy vegetable growth period identification and growth information prediction and analysis, and performs intelligent environmental control based on the detection results, thereby ensuring the accuracy and scientific nature of the strategy, being able to achieve on-demand lighting, reducing the waste of light energy and the "burned edge" phenomenon caused by excessive lighting, improving the yield and quality of leafy vegetables, and achieving better economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a structural schematic diagram of the regulating system in the present invention;

[0041] Figure 2 It is a structural schematic diagram of a growth period identification model in a growth detection module of the present invention;

[0042] Figure 3 It is a structural schematic diagram of a growth parameter prediction model in a growth detection module of the present invention;

[0043] Figure 4 It is a schematic diagram of the process of the adjustment method in the present invention;

[0044] In the attached figure, 1-leafy vegetables; 2-digital binocular camera; 3-sensor; 4-background management module; 5-growth detection module; 6-data storage module; 7-intelligent control module. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0046] In order to overcome the above-mentioned existing problems of low efficiency of manual detection and poor supplementary light control effect caused by reliance on experience in environmental control, a plant factory lighting control system and method based on intelligent identification of growth period is provided.

[0047] This embodiment relates to a plant factory lighting system based on intelligent identification of growth periods, such as Figure 1 As shown, this embodiment provides a plant factory lighting adjustment technology based on intelligent identification of growth periods. The system includes a background management module 4, a growth detection module 5, a data storage module 6 and an intelligent control module 7.

[0048] The system collects growth images and environmental parameter data of leafy vegetables in real time, and uses the background management module 4 to monitor and process the collected leafy vegetable growth information in real time. The growth detection module 5 predicts and calculates the growth information of the collected leafy vegetable growth RGB and depth images, and matches the supplementary strategy according to the staged supplementary lighting formula configured in the data storage module 6 (i.e., the database), adjusts environmental variables such as light, temperature and humidity, so as to achieve on-demand supplementary lighting, save energy while ensuring the yield and quality of leafy vegetables.

[0049] Backstage management module 4 obtains real-time RGB images and depth images of leafy vegetable growth through the EZVIZ cloud video protocol. In the backstage web interface, one can view the real-time images of digital binocular cameras at various locations on the production site, make manual and automatic adjustments to the fill light strategy, obtain real-time / historical data information of each sensor, and control the switch status of the actuator, parameter settings and other commands.

[0050] The growth detection module 5 includes a growth period recognition model and a growth parameter prediction model. The acquired growth RGB image is input into the growth period recognition model to obtain the growth period of the leafy vegetable; the acquired growth depth image and RGB image are input into the growth parameter prediction model to obtain the growth information of the leafy vegetable such as leaf area, stem height, dry and wet weight.

[0051] The growth period recognition model is an improved YOLO model, which includes a backbone network, a feature fusion network, and a prediction head. The backbone network extracts input image features; the feature fusion network fuses features of different scales; and the prediction head module outputs the growth period recognition results of the bounding box position and category label, such as Figure 2As shown in the figure. The C2F (Faster Implementation of CSP Bottleneck with 2convolutions) convolution module is introduced in the backbone network to reduce the amount of calculation while enhancing the feature extraction ability of the model. At the same time, the original PAFPN (Path Aggregation Feature Pyramid Network) structure is replaced with the more effective BiFPN (Bi-directional Feature Pyramid Network) structure in the feature fusion network to improve the network's fusion utilization of feature information of different scales. The self-attention mechanism CTF module is added to the prediction head, and the Transformer encoder structure is introduced to calculate the query matrix between different channels for the input features to further improve the perception of leafy vegetable image details. For example, when a 3×640×640 image is input, the CSPDarknet (Cross Stage Partial Network Darknet) backbone network generates feature maps of three scales, with sizes of 256×80×80, 512×40×40 and 1024×20×20 respectively. After the feature fusion network BiFPN, the output feature maps of the three scales fully integrate the information of feature maps of other scales. Finally, the global features are extracted from the Transformer encoder, and the decoder receives the encoder output and object query, and outputs the target features through a multi-layer attention mechanism. Finally, the target category and bounding box information are output through a one-dimensional convolution calculation.

[0052] Among them, the growth parameter prediction model is an improved multimodal image fusion network CDDFuse (Correlation-Driven feature Decomposition Fusion), including an Encoder module, a multimodal fusion module and a Decoder module, such as Figure 3As shown in the figure. The Encoder module consists of three components, namely, the Shared Feature Encoder (SFE) based on the Restormer block, the Base Transformer Encoder (BTE) based on the Transformer, and the Detail CNN Encoder (DCE) based on the CNN. The Encoder module is used to extract feature information from the image; the multimodal fusion module fuses the RGB features and the Depth features; the Decoder module outputs the predicted growth information such as leaf area, stem height, and dry and wet weight of leafy vegetables after decoding. In addition to the Transformer extraction channel, the Encoder module adds a CNN network channel to further extract feature information from the image, and finally superimposes the information extracted by the two channels; the SFE block in the Encoder module can extract shared features from high-resolution images by applying self-attention on the feature dimension, thereby extracting cross-modal shallow features without increasing too much computation; the BTE block uses a Transformer with spatial self-attention to extract long-distance dependent features; the CNN architecture in the DCE block can retain more detail information such as image texture edges and detail features, thereby extracting high-frequency detail information from shared features. The multimodal feature fusion module fuses the RGB image features and the depth image features into multimodal image features. After multimodal fusion, new image features are generated and input into the Decoder module. In the Decoder module, the Restormer block is used as the basic unit of the decoder to further improve the prediction ability of growth information, and finally output the predicted leaf area, stem height, dry and wet weight of leafy vegetables, etc. Figure 3 shown.

[0053] The data storage module 6 realizes persistent storage of the production control information data of the plant factory based on the SQL Server database, and stores the installation information, collected data and execution record information of the sensor equipment, actuators, digital binocular cameras and other devices installed in the plant factory in the cloud, which is convenient for the management and operation of production personnel. At the same time, the fill light formula and control strategy data are configured through the database, which also provides more flexibility for the regulation of plant factory equipment.

[0054] The intelligent control module 7 is divided into manual fill light mode and automatic fill light mode according to the control strategy set by the background management module 4. In the manual fill light mode, the fill light strategy can be manually selected to send commands to control the actuator to adjust. In the automatic fill light mode, the collected leaf vegetable images, sensor data and other information are passed through the growth detection module and then automatically select the best fill light strategy according to the fill light strategy in the database and control the actuator to adjust the lighting environment. Production personnel can log in to the system background to realize remote control.

[0055] The principle of the whole system: The system collects images and environmental data of the growth of leafy vegetables 1 in real time from plant factories or other agricultural production environments through digital binocular cameras 2 and sensors 3. After the image data is read by the background management software 4, the growth detection module 5 uses the improved YOLO model to identify the growth period, and then predicts the growth information through the improved multimodal image fusion network CDDFuse (Correlation-Driven feature Decomposition Fusion) to obtain growth data such as growth stage, leaf area, stem height, dry and wet weight, and save the growth data to the cloud SQL Server database 6. Then, the best fill light strategy is obtained by the pre-configured phased fill light formula and pre-set fill light strategy in the database. Finally, the intelligent control module 7 dynamically adjusts the actuator parameters such as the voltage, red and blue light ratio and light cycle of the LED lamp, and then dynamically adjusts the lighting environment to optimize the growth environment of leafy vegetables.

[0056] This embodiment also relates to a plant factory lighting method based on intelligent identification of growth periods. Taking leafy vegetables widely grown in plant factories as an example, the process is as follows: Figure 4 As shown, the following steps are included:

[0057] Step 1: Collect leafy vegetable images and environmental data in real time as basic information for system analysis.

[0058] For leafy vegetables grown in plant factories, a digital binocular camera 2 is installed to collect image data in real time. Environmental data is collected in real time by sensors, including key environmental parameters such as temperature, humidity, and light intensity. Leafy vegetable growth indicators such as leaf area index and plant height are obtained through measurement.

[0059] Step 2, the background management module 4 monitors the growth status of leafy vegetables in real time and manages the production remotely.

[0060] The real-time RGB and depth images of leafy vegetable growth are acquired through the EZVIZ Cloud video protocol, and then used as input to get the predicted growth information through the deployed growth detection algorithm. The backend web interface can be used to view the real-time images of the digital binocular cameras at various locations on the production site, perform manual and automatic adjustments to the fill light strategy, obtain real-time and historical data information of each sensor, and control the switch status of the actuator, parameter settings and other commands.

[0061] Step 3, the growth detection module 5 intelligently identifies the growth period of the leafy vegetables based on the image and predicts the growth information of the leafy vegetables.

[0062] The growth detection module includes a growth period recognition module based on the improved YOLO algorithm (such as Figure 2 ) and the growth parameter prediction module based on the improved CCDFuse algorithm (such as Figure 3 ).

[0063] The input leaf vegetable image will enter the growth stage recognition module. After the feature extraction in the backbone network, the feature maps of the three scales are fully fused in the feature fusion network. In the prediction head, the Transformer encoder structure is used to further improve the perception of the details of the leaf vegetable image. Finally, the growth stage and bounding box information of the leaf vegetable are output after one-dimensional convolution calculation, such as Figure 3 On the other hand, the RGB image and depth image of leafy vegetables are input into the growth parameter prediction module, and the multimodal image feature fusion is performed through the improved CDDFuse model. After multimodal fusion, new image features are generated and finally the predicted leaf area, stem height, dry and wet weight and other growth information of leafy vegetables are output, as shown in Figure 3 shown.

[0064] Step 4: the data storage module 6 stores the phased fill light formula and control strategy to provide a basis for intelligent control.

[0065] The leafy vegetable growth information obtained after processing by the growth detection module 5 will be stored in the data storage module 6, and matched according to the pre-configured stage fill light formula and control strategy to obtain the best fill light strategy under the current growth state, which will be input into the intelligent control module 7 as an input parameter.

[0066] Step 5: The intelligent control module 7 adjusts the operating parameters of the actuator based on the optimal parameters matched by the database.

[0067] The actuators (8) in the plant factory include humidifiers, air conditioners, LED lights, ventilation equipment, etc., which are responsible for dynamically adjusting the environmental parameters of leafy vegetables according to the generated control strategy. The intelligent control module 7 will adjust the operating parameters of the actuators according to the input parameters, such as adjusting the light intensity of the LED, the red and blue light ratio, and the light supplement cycle, so as to achieve on-demand light supplement and save energy.

[0068] The present invention realizes automatic recognition of leafy vegetable growth in a plant factory environment by constructing a recognition model through deep learning, and selects a lighting strategy suitable for leafy vegetable growth according to a carefully configured phased lighting formula, thereby achieving the purpose of on-demand lighting and energy saving.

[0069] The present invention combines computer vision technology with Web development technology to achieve intelligent management of leafy vegetables. By collecting leafy vegetable growth images and environmental data, using deep learning technology to detect and grade the growth status of leafy vegetables, formulating the optimal planting strategy, and intelligently adjusting the lighting parameters according to the strategy, it is possible to effectively reduce energy waste and the occurrence of "burned edges", improve the quality and yield of leafy vegetables, and at the same time reduce energy consumption, optimize resource utilization, and improve the economic benefits of planting. In practical applications, the present invention can not only meet the needs of leafy vegetable growth monitoring and environmental regulation, but also greatly reduce the reliance on manual experience, solve the problems of low efficiency and high cost in traditional agricultural production, promote the advancement of agricultural technology, and has broad application prospects and social value.

[0070] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A plant factory lighting adjustment system based on intelligent identification of growth period, characterized in that: The system comprises a background management module (4), a growth detection module (5), a data storage module (6) and an intelligent control module (7) which are connected in sequence; After receiving the real-time RGB image, depth image and environmental data of the leaf vegetable growth, the background management module (4) inputs the data into the growth detection module (5); The growth detection module (5) performs image preprocessing on the image, intelligently identifies the growth period of the current leafy vegetable, predicts the growth information of the current leafy vegetable, and sends it to the data storage module (6) for storage and recording; The data storage module (6) automatically matches the best lighting supplement strategy suitable for the current growth information of the leafy vegetables according to the pre-configured phased lighting supplement formula; The intelligent control module (7) dynamically adjusts the parameters of the actuator based on the optimal supplementary light strategy to maintain the optimal lighting environment for the leafy vegetables in real time.

2. According to claim 1, a plant factory lighting adjustment system based on intelligent identification of growth period is characterized in that: The growth detection module (5) includes a growth period recognition model and a growth parameter prediction model; The growth period recognition model is based on the RGB image of the leafy vegetable growth to identify the growth period of the leafy vegetable; The growth parameter prediction model predicts the growth information of leafy vegetables based on the depth image and RGB image of the leafy vegetable growth.

3. According to claim 2, a plant factory lighting adjustment system based on intelligent identification of growth period is characterized in that: The growth period recognition model is an improved YOLO model, including a backbone network, a feature fusion network and a prediction head; A C2F convolution module is introduced into the backbone network to extract image features. In the feature fusion network, the original PAFPN structure is replaced with a BiFPN structure to fuse feature information of different scales. A self-attention mechanism CTF module and a Transformer encoder structure are added to the prediction head to perceive the details of leafy vegetable images by calculating the query matrix between different channels.

4. According to claim 2, a plant factory lighting adjustment system based on intelligent identification of growth period is characterized in that: The growth parameter prediction model is an improved multimodal image fusion network, including an encoder module, a multimodal fusion module and a decoder module; The encoder module includes a shared feature encoder based on a Restormer block, and a Transformer-based basic encoder and a CNN-based detail encoder connected to the shared feature encoder; The multimodal fusion module is used to fuse the features output by the basic encoder and the detail encoder; The decoder module uses the Restormer block as the basic unit of the decoder.

5. A plant factory lighting adjustment system based on intelligent identification of growth period according to claim 4, characterized in that: The shared feature encoder extracts shared features from high-resolution images by applying self-attention on the feature dimension, thereby extracting cross-modal shallow features; The basic encoder uses a Transformer with spatial self-attention to extract long-range dependency features from the shared features output by the shared feature encoder; The detail encoder extracts high-frequency detail information from the shared features output by the shared feature encoder based on CNN.

6. The plant factory lighting adjustment system based on growth period intelligent identification according to claim 1 is characterized in that: The growth information of the leafy vegetables includes leaf area, stem height and dry and wet weight of the leafy vegetables.

7. The plant factory lighting adjustment system based on growth period intelligent identification according to claim 1 is characterized in that: The actuator parameters include the voltage, red and blue light ratio and photoperiod of the LED lamp.

8. The plant factory lighting adjustment system based on intelligent identification of growth period according to claim 1 is characterized in that: The system further comprises a digital binocular camera (2) and a sensor (3), wherein the digital binocular camera (2) is used to collect real-time images of the growth of leafy vegetables; and the sensor (3) is used to collect environmental data of the growth of leafy vegetables.

9. The plant factory lighting adjustment system based on growth period intelligent identification according to claim 1 is characterized in that: The background management module (4) monitors the growth status of the leafy vegetables in real time and controls the commands of the execution mechanism.

10. A method for adjusting the lighting adjustment system of a plant factory based on intelligent identification of growth periods according to any one of claims 1 to 9, characterized in that: The method comprises the following steps: Step 1, collecting images and environmental data of leafy vegetable growth in real time through digital binocular cameras and sensors; Step 2, the background management module (4) monitors the growth status of leafy vegetables in real time and manages production remotely; Step 3, the growth detection module (5) intelligently identifies the growth period of the leafy vegetables based on the image, and predicts the growth information of the leafy vegetables; Step 4, the data storage module (6) automatically matches the best lighting strategy suitable for the current growth information of the leafy vegetables according to the pre-configured phased lighting formula; Step 5: The intelligent control module (7) controls the actuator to adjust the operating parameters based on the optimal fill light strategy.

Images