Intelligent accelerated breeding cabin and method for regulating and controlling environment in cabin

Abstract

The invention provides an intelligent accelerated breeding cabin, and belongs to the field of crop cultivation, the breeding cabin comprises an independent crop cultivation area and an equipment area, the crop cultivation area is provided with an image acquisition module, a detection module, a regulation and control module and at least one tray rack, and the equipment area is provided with a visual identification module, an upper computer and a controller; the image acquisition module acquires growth images of crops on the tray rack and transmits the growth images to the visual identification module, the visual identification module predicts the growth stage of the crops and transmits the growth stage to the upper computer, and the upper computer generates a control instruction according to target parameters corresponding to the growth stage of the crops and transmits the control instruction to the controller. The detection module detects in-cabin environment parameters in real time and feeds back the parameters to the controller, and the controller drives the regulation and control module to regulate the in-cabin environment parameters until target parameters are reached; the invention further provides a method for regulating and controlling the environment in the breeding cabin. A cabin environment regulation and control mechanism of visual identification-decision control-closed loop feedback is formed, and precise regulation and control of the cabin environment according to dynamic requirements of the cabin crop growth stage is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of crop cultivation, and particularly to an intelligent accelerated breeding cabin and a method for regulating the environment inside the cabin. Background Art

[0002] In today's society, due to multiple factors such as the continuous growth of the population and the continuous fluctuations of the climate, the problem of crop cultivation has attracted increasing attention. With the increasing improvement of people's living standards, the demand for food has increased sharply, and more stringent requirements have been put forward for the nutritional value and safety standards of food. Facing this challenge, cultivating high-yield and high-quality crops has become the key path to alleviate the contradiction between food supply and demand. However, using traditional breeding techniques to screen and cultivate excellent crop varieties often takes an extremely long time, which significantly hinders the rapid progress of breeding work. Therefore, using appropriate technical means to create the best growth conditions for crops, thereby promoting the growth rate of crops, shortening their growth cycle, and increasing the iteration rate, has become the core focus of current accelerated breeding research. Thus, the existing greenhouse cultivation technology has emerged, but the environmental regulation of traditional greenhouses or plant factories mainly relies on artificial experience or fixed parameter settings and cannot be accurately regulated according to the dynamic needs of the crop growth stage.

[0003] In the prior art, the Chinese patent application for invention "An Intelligent Plant Generation Accelerated Breeding Cabin" with the publication number CN111296129A discloses an intelligent plant generation accelerated breeding cabin, which includes a cabin body. The cabin body is provided with a control system area for controlling all equipment in the cabin, a planting area for crop planting operations, and an equipment placement area for placing planting equipment. The planting area is provided with multiple groups and multiple layers of planting modules and environmental parameter detection devices. The environmental parameter detection devices are used to intelligently detect the environmental parameters of the planting area, monitor the data in real time, and feedback and timely adjust the environmental parameters required by plants through an intelligent control terminal. The intelligent control terminal controls the air conditioner within the range of 18°C - 40°C and controls the ultrasonic humidifier within the range of 60% - 80% RH. It can be seen that the environmental regulation of this breeding cabin is mainly based on fixed parameter settings, and the growth stage of crops cannot be monitored, and dynamic optimization cannot be performed according to the real-time growth state of crops. Summary of the Invention

[0004] The technical problem to be solved by the present invention is how to solve the problem that the existing breeding cabin cannot accurately regulate the environment inside the cabin according to the dynamic needs of the crop growth stage.

[0005] The present invention solves the above technical problems through the following technical solutions: an intelligent accelerated breeding cabin, which includes an independent crop cultivation area and an equipment area. The crop cultivation area is provided with an image acquisition module, a detection module, a regulation module, and at least one tray rack. The equipment area is provided with a visual recognition module, a host computer, and a controller. The image acquisition module acquires the growth images of the crops on the tray rack and transmits them to the visual recognition module. The visual recognition module predicts the crop growth stage and transmits it to the host computer. The host computer generates a control instruction according to the target parameters corresponding to the crop growth stage and transmits it to the controller. The detection module continuously detects the environmental parameters in the cabin and feeds them back to the controller. The controller drives the regulation module to adjust the environmental parameters in the cabin until the target parameters are reached.

[0006] Beneficial effects: By setting an image acquisition module in the breeding cabin, the present invention realizes the real-time acquisition of crop growth images. The acquired images are input into the visual recognition module, and the growth stage of the crops is mapped. The host computer generates a control instruction according to the target parameters of the growth environment required for the crop growth stage. After comparing the control instruction with the current environmental parameters continuously collected by the detection module, the controller performs closed-loop control on the environmental parameters in the cabin, forming an environmental regulation mechanism of "visual recognition - decision control - closed-loop feedback" in the cabin, realizing precise regulation of the environmental parameters in the cabin according to the dynamic requirements of the crop growth stage in the cabin, and further realizing the precision and automatic control of crop growth management.

[0007] Preferably, the visual recognition module is built-in with a trained YOLOv8 model. The training process of the trained YOLOv8 model is as follows:

[0008] Every day, the growth images of the crops in the cabin are continuously captured in a combination of vertical downward shooting and side shooting. The images are labeled with bounding boxes and growth stage categories, the labeled images are enhanced, and the growth stage categories are balanced to construct a data set;

[0009] The YOLOv8 model is initialized, with the image as the input and the crop growth stage as the output, and the YOLOv8 model is trained. When the set number of training rounds is reached, the trained YOLOv8 model is obtained.

[0010] Beneficial effects: The visual recognition module of the present invention uses the YOLOv8 deep learning model to accurately identify the crop growth stage, combines CBAM (Convolutional Block Attention Module) to improve the ability to capture leaf edge and texture features, and the data enhancement strategy and category balance method for the crop growth stage enable the vision system to achieve a very high accuracy rate in the recognition of the crop growth stage.

[0011] Preferably, the detection module includes a temperature sensor, a humidity sensor, a CO2 concentration sensor, a photoelectric sensor, and an illuminance sensor, all of which are connected to the controller. The regulation module includes a refrigerator, a heater, a dehumidifier, a spray valve, a fan, and a motor, all of which are also connected to the controller. The detection module collects the current temperature, humidity, CO2 concentration, and illuminance and transmits them to the controller. The controller calculates the control signal based on the difference between the current temperature, humidity, CO2 concentration, illuminance and the target parameters, and sends it to the regulation module to respectively control the start and stop of the refrigerator and the heater, the start and stop of the dehumidifier and the spray valve, the start and stop of the solenoid valve and the fan, and the rise and fall of the motor, so as to adjust the temperature, humidity, CO2 concentration, and illuminance in the cabin.

[0012] Beneficial effects: The present invention combines visual recognition with multi-sensors, breaks through the limitation of traditional breeding systems relying on manual observation, solves the problem of lag in environmental adjustment during unattended operation at night. By real-time monitoring the plant growth status and environmental parameters, the system can automatically adjust factors such as temperature, humidity, CO2 concentration, and light in the breeding cabin, realizing unmanned management, saving labor and material costs, and improving breeding efficiency and quality.

[0013] Preferably, the controller is a PLC with a built-in PID regulator, and the parameters of the PID regulator are adjusted using an improved particle swarm optimization algorithm, with the ITAE evaluation index as the fitness function f i :

[0014]

[0015] where e(t) is the system error, t is the system running time, and the calculation formulas for the learning factors c1 and c2 are:

[0016]

[0017] where c 1_max , c 1_min , c 2_max , c 2_min are the upper and lower limits of the learning factors c1 and c2 respectively.

[0018] Beneficial effects: The present invention adopts non-linear learning factors c1 and c2 constructed based on sine functions, enabling them to change dynamically and adapt to the environment during the iteration process, thus solving the problem that fixed learning factors show a decline in convergence during the iteration process, reducing the global search ability of the algorithm and being prone to falling into local optima.

[0019] Preferably, the adaptive inertia weight factor ω(t) of the improved particle swarm optimization algorithm is:

[0020]

[0021] Among them, τ is the weight decreasing adjustment factor.

[0022] Beneficial effects: The present invention uses an adaptive inertia weight factor, which can be dynamically adjusted according to the search stage. By improving the adaptive mechanisms of the learning factor and the weight coefficient, the flexibility of PSO in different optimization stages is enhanced. Through the combination of the improved particle swarm optimization algorithm and the PID controller for dynamic optimization regulation, the temperature fluctuation is controlled within ±0.5°C, and the humidity fluctuation is within ±3%.

[0023] Preferably, the temperature sensor, the humidity sensor, and the CO2 concentration sensor are respectively located on the side of the tray rack, the photoelectric sensor and the illuminance sensor are respectively located above the tray rack, the refrigerator is located in the equipment room, the heater, the dehumidifier, and the fan are respectively located on both side walls of the breeding cabin. Heat outlets and refrigerator air inlets are respectively provided on both side walls of the breeding cabin, a fresh air inlet is provided on one side wall of the breeding cabin, a circulating air duct is provided in the crop cultivation area, the circulating air duct is communicated with the CO2 air outlet on the top wall of the breeding cabin, and a humidification pipeline communicated with the spray valve is provided near the tray rack.

[0024] Beneficial effects: The internal layout of the breeding cabin of the present invention is reasonably set and highly integrated. Considering that most of the electrical energy of the light source is converted into heat and dissipated into the cabin environment, a cold air blower is used to continuously adjust the temperature through the refrigerator air inlet, and a circulating fan is used to keep the air in the cabin flowing, making the temperature and humidity in the cabin more uniform.

[0025] Preferably, a multi-spectral light source installed on a lifting lamp rack is provided above the tray rack, and the lifting lamp rack is fixedly installed on the top wall of the breeding cabin.

[0026] Beneficial effects: The multi-spectral light source adopted by the light source of the present invention has the advantages of small heat generation, precise and adjustable light formula, diverse installation adaptation modes, long service life, etc., and can also be flexibly modulated according to the growth characteristics and production purposes of crops.

[0027] Preferably, the breeding cabin further includes a cultivation rack located on the tray rack, a cultivation tank located on the cultivation rack, and a water and fertilizer circulation system communicated with the cultivation tank. The water and fertilizer circulation system includes a water addition pipeline, and the controller controls the electromagnetic valve and the water pump on the water addition pipeline to inject nutrient solution into the cultivation tank.

[0028] Beneficial effects: By setting the water and fertilizer circulation system, the present invention can meet the needs of hydroponic plants or soil-cultivated plants at the same time.

[0029] Preferably, the breeding cabin further includes a cabin body and an intelligent electric door installed in the crop cultivation area, and a single-opening door is installed in the equipment area.

[0030] The present invention also provides a method for regulating the cabin environment, which is applied to the intelligent accelerated breeding cabin, and the method includes:

[0031] Collect the crop growth images in the collection cabin, preprocess the images, and obtain the preprocessed images;

[0032] Input the preprocessed images into the visual recognition module, predict the crop growth stage, and transmit it to the host computer;

[0033] The host computer generates control instructions according to the target parameters corresponding to the crop growth stage and transmits them to the controller;

[0034] The detection module continuously detects the environmental parameters in the cabin and feeds them back to the controller. The controller drives the regulation module to adjust the environmental parameters in the cabin until the target parameters are reached.

[0035] The advantages provided by the present invention are as follows:

[0036] The present invention dynamically adjusts the environmental parameters (such as temperature, humidity, light intensity, CO2 concentration) according to the real-time recognition results of the crop growth stage, adopts the PID control algorithm and the reinforcement learning optimization strategy to achieve closed-loop feedback regulation, ensures the stability of the environmental parameters (temperature fluctuation ±0.5°C, humidity fluctuation ±3%), provides an anomaly detection and fault redundancy mechanism to ensure the reliable operation of the system. Based on the visual recognition technology, the control of the entire cabin environment is completed, saving the resources of manual supervision, improving the traditional control efficiency, and exceeding the efficiency of traditional manual intervention in terms of time response efficiency. Brief Description of the Drawings

[0037] Figure 1 It is a three-dimensional view of the intelligent accelerated breeding cabin provided by the embodiment of the present invention;

[0038] Figure 2 It is a three-dimensional view of the intelligent accelerated breeding cabin provided by the embodiment of the present invention;

[0039] Figure 3 It is the layout diagram inside the intelligent accelerated breeding cabin provided by the embodiment of the present invention;

[0040] Figure 4 It is the schematic diagram of the control principle of the environmental parameters inside the intelligent accelerated breeding cabin provided by the embodiment of the present invention;

[0041] Figure 5 It is the distribution diagram of the monitoring parameters and control functions of the intelligent accelerated breeding cabin provided by the embodiment of the present invention;

[0042] Figure 6 It is the fitness curve of the improved PSO algorithm adopted by the controller in the intelligent accelerated breeding cabin provided by the embodiment of the present invention;

[0043] Figure 7 It is the flow chart of the method for regulating the environment inside the intelligent accelerated breeding cabin provided by the embodiment of the present invention. Detailed Embodiments

[0044] To make the objectives, technical solutions, and advantages of the present invention more clear and understandable, the following describes the technical solutions of the present invention clearly and completely in combination with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts fall within the scope of protection of the present invention.

[0045] As Figure 1 shown, this embodiment provides an intelligent accelerated breeding cabin. The breeding cabin includes an independent crop cultivation area 1 and an equipment area 2. The crop cultivation area 1 is equipped with an image acquisition module, a detection module, a regulation module, and at least one pallet rack 10. Figure 2 As shown in the breeding cabin, two pallet racks 10 are configured. Pallets are placed on the pallet racks 10, and cultivation racks are placed on the pallets. The material of the cultivation racks is corrosion-resistant. The cultivation racks can be assembled and the layer spacing can be adjusted, which is convenient for replacing cultivation varieties. Cultivation grooves are placed on the cultivation racks for holding nutrient solution or soil. When cultivating crops, the crops are inserted into the cultivation grooves. Each pallet can be transported independently to achieve the transportation and phenotypic detection of crops. When transporting the pallet, first adjust the forklift to the front of the breeding cabin, then lift it about 230 mm and insert it into the middle of the pallet. Then lift the pallet smoothly by at least 80 mm to cross the limit block. Finally, the forklift slowly retreats to completely separate the pallet from the cabin body and then gently puts it down. When putting in the pallet, pay attention to the pallet limit block. The best distance between the pallet support feet and the limit block is 5 mm.

[0046] The equipment area 2 is equipped with a visual recognition module, a host computer, and a controller. The image acquisition module acquires the growth images of the crops on the pallet rack 10 and transmits them to the visual recognition module. The visual recognition module predicts the crop growth stage and transmits it to the host computer. The host computer generates a control instruction according to the target parameters corresponding to the crop growth stage and transmits it to the controller. The detection module continuously detects the environmental parameters in the cabin and feeds them back to the controller. The controller drives the regulation module to adjust the environmental parameters in the cabin until the target parameters are reached.

[0047] The image acquisition module is a high-resolution camera (RGB, multispectral, or thermal imaging) for acquiring crop image data. A database is established for different growth stages, covering key growth stages such as the germination stage, tillering stage, heading stage, and maturity stage. It shoots continuously for 22 hours every day in a combination of vertical top-down shooting and 45-degree side shooting to ensure stable light and no shadow interference. The camera has the ability of automatic zoom. The camera can be connected to a computer through Ethernet, and the computer can directly log in to the camera through the IP address to view the video situation.

[0048] Use the LabelImg tool to annotate the collected crop images with bounding boxes and growth stages. When annotating the data, key features (such as plant height, number of leaves, color change) need to be included. Construct a standardized dataset containing 10,000 images, and perform enhancement strategies such as random rotation, brightness adjustment, and affine transformation on the annotated images to improve diversity. At the same time, use oversampling technology to balance the categories and ensure that the sample differences in each stage do not exceed 10%, obtaining the dataset.

[0049] The visual recognition module is trained and optimized based on the YOLOv8 model. The model is initialized with pre-trained weights (yolov8n.pt). Configure the PyTorch 2.0 and CUDA 11.8 environments under the Ubuntu 22.04 system, and use the NVIDIA RTX4080 GPU to accelerate the training. The training hyperparameters are set as follows: initial learning rate 0.01, momentum 0.937, batch size 16, number of training epochs 300, input image size 640×640. To further improve the model's ability to capture crop details, embed CBAM (Convolutional Block Attention Module) in the Backbone to enhance the extraction of leaf edge and texture features; at the same time, fuse multi-spectral data (near-infrared band), and reduce the number of channels to 3 dimensions through 1×1 convolution to achieve multi-modal fusion of RGB and physiological features. Train the YOLOv8 model based on the dataset, and when the loss function is minimized, obtain the prediction unit. The final model achieves an mAP@0.5 accuracy of 94.2% on the test set, a single-frame inference speed of 45 FPS, and an F1-Score of 92.8%.

[0050] After completing the model training, convert it to the ONNX format and deploy it to the NVIDIA Jetson AGX Orin edge computing device to achieve real-time inference through TensorRT acceleration. The visual recognition results (in JSON format) are transmitted to the host computer developed based on C++ and QT via the TCP / IP protocol to display the crop growth stage, environmental parameters, and device status in real time. The determination of the growth stage is realized through feature extraction, using PyTorch to extract morphological features (leaf area index, stem diameter) and color histograms. Perform LSTM / 3D-CNN modeling on the time-series images to capture the dynamic growth law. Output the probability distribution in the stage classification (such as "heading stage" confidence 85%) to avoid the risk of misjudgment of a single label.

[0051] A host computer is designed based on the C++ language and QT. The host computer receives the visual analysis results, obtains the crop growth stage data from the visual analysis module, generates control instructions according to preset rules (such as 25°C and 60% humidity during the heading stage), communicates with the PLC through the Modbus TCP protocol, and drives actuators such as cooling fans, heaters, and humidifiers to form a closed-loop regulation. For example, when the "heading stage" is detected, the host computer sends an instruction to the PLC to adjust the red / blue light ratio of the LED light source to 3:1 and increase the CO2 concentration to 800 ppm. The host computer software architecture developed based on C++ and QT supports real-time data display, device control, historical data playback, and abnormal alarm functions. The communication interface design between the host computer and the visual recognition module and the PLC ensures efficient data transmission and accurate instruction execution.

[0052] This invention combines the visual recognition module with the in-cabin environment regulation module to form a closed-loop feedback system, realizing the full-process automation from crop growth stage recognition to dynamic adjustment of environmental parameters. For the environmental parameter mapping table for different crop growth stages (such as the set values of temperature, humidity, and light intensity during the germination stage, tillering stage, and heading stage), after being processed by the host computer, instructions are sent to control the controller to adjust the in-cabin environment. The visual recognition module of this invention is built with the trained YOLOv8 model. The YOLOv8 deep learning model is used to accurately identify the crop growth stage. Combining with the CBAM (Convolutional Block Attention Module) improves the ability to capture leaf edge and texture features. The data augmentation strategy (such as random rotation, brightness adjustment, and affine transformation) and class balancing method (oversampling technique) for crop growth stage data enable the visual system to achieve a very high accuracy in the identification of crop growth stages.

[0053] See Figure 2 and Figure 3 , the detection module includes a temperature sensor, a humidity sensor, a CO2 concentration sensor, a photoelectric sensor, and a illuminance sensor 30 respectively connected to the controller. The regulation module includes a refrigerator, a heater, a dehumidifier, a spray valve, a fan, and a motor respectively connected to the controller. The detection module collects the current temperature, humidity, CO2 concentration, and illuminance and transmits them to the controller. The controller calculates the control signal based on the difference between the current temperature, humidity, CO2 concentration, and illuminance and the target parameters, and sends it to the regulation module to respectively control the start and stop of the refrigerator and the heater, the start and stop of the dehumidifier and the spray valve, the start and stop of the solenoid valve and the fan, and the rise and fall of the motor, and adjusts the temperature, humidity, CO2 concentration, and illuminance in the cabin.

[0054] The temperature sensor collects the current temperature inside the cabin and transmits it to the controller. The controller calculates the control signal based on the difference between the current temperature and the target temperature, and sends it to the refrigerator or heater to adjust the air outlet of the refrigerator and the hot air outlet to achieve precise control of the temperature inside the cabin within the range of 15°C - 40°C, so as to balance the temperature in the area of the pallet rack 10. The humidity sensor collects the current humidity inside the cabin and transmits it to the controller. The controller calculates the control signal based on the difference between the current humidity and the target humidity, and sends it to the dehumidifier or spray valve to balance the humidity in the area of the plant pallet rack 10 through precise control of the humidification pipeline. The CO2 concentration sensor collects the current CO2 concentration inside the cabin and transmits it to the controller. The controller calculates the control signal based on the difference between the current CO2 concentration and the target concentration, and sends it to the fan. By controlling the timing switch of the fresh air outlet, the air environment inside the cabin can always meet the air environment required by the plants and the demand for air fluidity by the plants, thereby promoting the growth of the plants. The illuminance sensor 30 collects the current illuminance inside the cabin and transmits it to the controller. The controller calculates the control signal based on the difference between the current illuminance and the target illuminance, and sends it to the motor. By precisely adjusting the multi-spectral LED chip combination light source on the lifting lamp rack and adjusting the height of the lifting lamp rack, the components of the light quality, light intensity, and light time required by the plants can be met, enabling the plants to always be under suitable lighting conditions, thereby promoting the growth of the plants.

[0055] The controller is a PLC with a built-in PID regulator. The model of the PLC is XD5E-32T. The control method of combining the improved particle swarm optimization algorithm and the PID controller is used for integrated control of the cabin environment:

[0056] In this paper, the optimized dimension of the algorithm is three-dimensional, corresponding to the three parameters of the PID controller. ITAE is used to weight the objective function, the population size is 100, and the maximum number of iterations is 100.

[0057] Through the fitness function, the fitness value of each particle can be obtained, and then the individual best and the global best can be compared. Taking the ITAE evaluation index as the fitness function f of the PLC parameters i , the comprehensive performance of the system is evaluated by the absolute integral of the particle time, and the expression is:

[0058]

[0059] Among them, e(t) is the system error, and t is the system operation time.

[0060] In the traditional particle swarm optimization algorithm, the learning factor c1 remains unchanged during the iteration process. The fixed learning factor shows a decline in convergence during the iteration process, reducing the global search ability of the algorithm and making it easy to fall into local optima. To solve this problem, the present invention adopts non-linear learning factors c1 and c2 constructed based on the sine function, enabling them to change dynamically and adapt to the environment during the iteration process. The update process is as follows:

[0061]

[0062] where c 1_max , c 1_min are the upper and lower limits of the learning factor c1 respectively, and c 2_max , c 2_min are the upper and lower limits of the learning factor c2 respectively.

[0063] In addition, the inertia weight factor in the PSO algorithm is used to regulate the ability of the particle to inherit the velocity of the previous iteration. Generally, a larger inertia weight factor is required at the initial stage of the search to achieve global optimization, and a smaller inertia weight factor is required at the later stage of the search to strengthen the local search ability. The present invention uses an adaptive inertia weight factor, which can be dynamically adjusted according to the search stage. The adaptive inertia weight factor ω(t) is:

[0064]

[0065] where τ is the weight decay adjustment factor, and τ can be taken as τ = 2.0.

[0066] In the improved PSO algorithm, when the initial position distribution of the particles is more uniform, by improving the adaptive mechanism of the learning factor and the weight coefficient, the flexibility of the PSO in different optimization stages is enhanced. The fitness curve of the algorithm is as Figure 6 shown. The final ITAE of the obtained system is 0.00317, the overshoot is 0.114%, and the adjustment time is 0.083 s. From the comprehensive test results, it can be concluded that the proposed improved PSO algorithm has significant advantages in the optimization problems of this type of system, and the test results fully verify the effectiveness and superiority of this method.

[0067] Through the integrated design of "visual recognition - decision control - closed-loop feedback", the present invention realizes the precision and automatic control of crop growth management, provides efficient and reliable technical support for intelligent plant factories and precision agriculture. By improving the particle swarm optimization algorithm combined with a PID controller for dynamic optimization regulation, the temperature fluctuation is controlled within ±0.5°C, and the humidity fluctuation is within ±3%. The test of continuous operation for 30 days in the tomato planting cabin shows that the model recognition accuracy reaches 93.5%, and the misjudgment mainly focuses on the transition period of the growth stage (such as the tillering stage to the jointing stage). Compared with traditional planting, the system energy consumption is reduced by 40% (the average daily power consumption is 9 kWh), the crop growth cycle is shortened by 18%, and the yield is increased by 22%. In addition, the QT upper computer provides a visual interface, supporting functions such as real-time monitoring, historical data playback, and abnormal alarm, significantly reducing the need for manual intervention. The integrated hardware design integrates an integrated hardware design solution for cameras, sensors, and actuators, reducing the system complexity and deployment cost. The scalable modular design supports the rapid addition of new functions (such as the pest and disease detection module).

[0068] The temperature sensor, humidity sensor, and CO2 concentration sensor are respectively located on the side of the tray rack, such as Figure 2 in the area 20 shown. The photoelectric sensor and the illuminance sensor 30 are respectively located above the tray rack 10. The refrigerator is located in the equipment room. The heater, dehumidifier, and fan are respectively located on both side walls of the breeding cabin. Heat outlets and refrigerator air outlets are respectively opened on both side walls 50 of the breeding cabin. A fresh air outlet is opened on one side wall of the breeding cabin. A circulation air duct is provided in the crop cultivation area 1, and the circulation air duct is communicated with the CO2 air outlet on the top wall of the breeding cabin. A humidification pipeline communicated with the spray valve is provided near the tray rack. Valves and water pumps are provided on the humidification pipeline. The humidification process is operated in cooperation with the valves and water pumps. It is necessary to first open the water inlet solenoid valve, and then turn on the water inlet pump. After humidification, first turn off the water inlet pump, and then close the humidification solenoid valve.

[0069] The breeding cabin also includes a water and fertilizer circulation system communicated with the cultivation tank. The water and fertilizer circulation system includes a water addition pipeline. Solenoid valves and water pumps are provided on the water addition pipeline. The controller injects nutrient solution into the cultivation tank by controlling the solenoid valves and water pumps on the water addition pipeline to meet the irrigation requirements of the crops. If the crop is a hydroponic plant, the water addition or drainage of the cultivation tank is realized by controlling the solenoid valves and water pumps. The drainage process is operated in cooperation with the solenoid valves and water pumps. During the drainage process, first open the solenoid valve, and then turn on the drainage pump. When the drainage ends, first turn off the drainage pump, and then close the solenoid valve.

[0070] Above the pallet rack 10, there is a multi-spectral light source installed on the lifting lamp rack 60. The lifting lamp rack 60 is fixedly installed on the top wall of the breeding cabin. As the energy source and important signal source for plant growth, the artificial light source is a crucial part of the breeding cabin design. In this invention, the multi-spectral LED chip combined light source is adopted on the lifting lamp rack as the artificial light source, which has the advantages of small heat generation, precise and adjustable light formula, diverse installation and adaptation modes, long service life, etc. Moreover, it can be flexibly modulated according to the growth characteristics and production purposes of crops. For the multi-light quality chip combined LED light source, attention should be paid to the arrangement density and light-emitting angle of each lamp bead to ensure as much as possible the uniform distribution of the mixed light quality light on the lower illumination surface. Although the photoelectric conversion efficiency of the LED light source has been greatly improved compared with traditional light sources, a large part of the electrical energy is still converted into heat and dissipated into the cabin environment. Therefore, it is necessary to use a cold air blower to continuously adjust the temperature through the refrigeration machine air outlet. In order to make the temperature and humidity in the cabin uniform, it is necessary to use a circulation fan to keep the air in the cabin flowing. When designing, the appropriate circulation fan and its air supply and return methods need to be determined according to the scale of the cabin. Since a large part of the electrical energy consumption is used for environmental control, therefore, in the design, the application of energy-saving technologies needs to be fully considered. By selecting units with higher energy efficiency ratios, accurately calculating the cooling capacity, and reasonably selecting air-conditioning units, the energy saving of the system can be achieved. In addition, the time of CO2 fertilization and external circulation ventilation needs to be reasonably allocated to ensure a high resource utilization efficiency.

[0071] The breeding cabin also includes a cabin body and an intelligent electric door 90. The intelligent electric door 90 is installed in the crop cultivation area, and a single-opening door 80 is installed in the equipment area. An in-cabin status alarm indicator light 70 is installed on the top of the breeding cabin. The cabin body adopts a frame-type large-panel structure, which is divided into four layers, with an external dimension of 6000mm×2500mm×2500mm and a protection level of IP54. In order to maintain the set environmental parameters in the cabin, ensure the efficient generation of generations of crops, prevent the intrusion of external pests and diseases, and ensure the high quality and uniformity of products, the cabin body selects an opaque heat-insulating material to isolate the internal environment from the outside world. Therefore, a multi-layer three-dimensional modular cabin is designed. The paint color number of the outside paint spray is Iveco white (white), the outer skin is 1.2mm steel plate, and the inner skin is 0.8mm stainless steel, with a reasonable layout and heat preservation effect. The cabin body weighs 4.9t (including empty pallets). There is a total of 1 gas path in the cabin body. The gas path is made of 4-point seamless 304 stainless steel pipe, and the main body is embedded in the left side wall, and an opening is made on the left side wall to lead out the outlet.

[0072] The power supply of the intelligent electric door 90 is AC220V, and the mode is to trigger induction to open the door once and close the door once. When it is turned on for the first time, the locking mechanism of each door needs to be removed. The locking mechanism is located below the tail of the door body. The moving door system has 4 door limit switches, which are respectively placed on the upper side of the door, and the switch type is normally open. Through the automatic opening and closing of the intelligent electric door 90, it can interact with external robots to realize functions such as identification, grasping, transportation, and detection of crops to be tested.

[0073] The wiring on the right wall inside the cabin includes the door power supply, door induction wire, door limit wire, and 3-color light wire, which are routed through the 6030 wire trough, pass through the refrigeration internal unit channel, and run along the refrigeration machine pipeline. The wiring on the left wall includes the front heating pipeline, front fan wire, front heat dissipation fan wire, front monitoring wire, rear heating pipeline, rear fan wire, rear heat dissipation fan wire, and rear monitoring wire. They pass through the 7550 wire trough, go down along the vertical partition board, and then pass through the vertical partition board to enter the equipment cabin. The lighting wire of the equipment cabin passes through the refrigeration internal unit channel and runs along the refrigeration pipeline. The power supply of the cabin is AC380V, 50Hz, and the grounding point of the cabin is on the right side of the equipment cabin door. The standard cabin incoming line adopts the connector form, and the 3-phase 5-wire incoming line is input.

[0074] All the equipment inside the cabin is controlled by the PLC controller to start and stop. The human-machine interaction is a touch screen, and the touch screen is mainly divided into 5 operation interfaces, namely working condition operation, equipment control, cabin door control, lighting control, and parameter setting.

[0075] Working condition operation: Displays the operating status of all equipment, as well as the start and stop of equipment under the linkage system.

[0076] Equipment control: The equipment single-control interface, where each piece of equipment can be controlled separately in the interface.

[0077] Cabin door control: Controls the opening and closing of the cabin door. Before the cabin door is opened, it will be enabled to open. When it is enabled to open, the cabin door will perform a self-check operation. In the non-enabled state, the opening and closing operations are invalid.

[0078] Lighting control: Can control each road of plant growth lights and control the lifting frame.

[0079] Parameter setting: Displays parameters such as the current illuminance, temperature and humidity, and carbon dioxide concentration inside the cabin, and can set the corresponding parameters.

[0080] See Figure 7 , the present invention also provides a method for regulating the internal environment of an intelligent accelerated breeding cabin, including:

[0081] Step 1, collect the crop growth images inside the cabin, preprocess the images to obtain the preprocessed images;

[0082] Step 2, input the preprocessed images into the visual recognition module, predict the crop growth stage, and transmit it to the upper computer;

[0083] Step 3: The host computer generates a control instruction according to the target parameters corresponding to the crop growth stage and transmits it to the controller.

[0084] Step 4: The detection module continuously detects the environmental parameters in the cabin and feeds them back to the controller, and the controller drives the regulation module to adjust the environmental parameters in the cabin until the target parameters are reached.

[0085] Most of the existing environmental control systems are open-loop controls, relying on fixed parameter settings (such as timed switching of lights, constant temperature control), and cannot dynamically adjust according to the growth state of crops. Although some systems introduce sensor data (such as temperature, humidity, CO2 concentration), they lack in-depth integration with visual recognition, and the control accuracy is limited. The present invention dynamically adjusts environmental parameters (such as temperature, humidity, light intensity, CO2 concentration) according to the real-time recognition results of crop growth stages, adopts a PID control algorithm and a reinforcement learning optimization strategy to achieve closed-loop feedback control, and ensures the stability of environmental parameters (temperature fluctuation ±0.5°C, humidity fluctuation ±3%). Based on visual recognition technology, the control of the entire cabin environment is completed, saving the resources of manual guarding, improving the efficiency of traditional control, and exceeding the efficiency of traditional manual intervention in terms of time response efficiency. Integrate functions such as visual recognition, environmental control, and user interaction into a unified platform to achieve a highly integrated design. Deploy the model based on edge computing devices, reduce dependence on the cloud, reduce communication latency, provide an integrated hardware design solution, integrate cameras, sensors, and actuators, and reduce deployment costs and complexity.

[0086] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. Intelligent acceleration breeding cabin, characterized in that: The breeding cabin includes an independent crop cultivation area and an equipment area. The crop cultivation area is equipped with an image acquisition module, a detection module, a regulation module, and at least one tray rack. The equipment area is equipped with a visual recognition module, a host computer, and a controller. The image acquisition module captures the growth images of the crops on the tray rack and transmits them to the visual recognition module. The visual recognition module predicts the crop growth stage and transmits it to the host computer. The host computer generates a control instruction according to the target parameters corresponding to the crop growth stage and transmits it to the controller. The detection module continuously detects the environmental parameters in the cabin and feeds them back to the controller. The controller drives the regulation module to adjust the environmental parameters in the cabin until the target parameters are reached.

2. The intelligent accelerated breeding cabin according to claim 1, characterized in that: The visual recognition module is built-in with the trained YOLOv8 model. The training process of the trained YOLOv8 model is as follows: Every day, the growth images of the crops in the cabin are taken continuously in a combination of vertical top-down and side views. The images are annotated with bounding boxes and growth stage categories, enhanced, and the growth stage categories are balanced to construct a dataset. The YOLOv8 model is initialized, with images as input and crop growth stages as output, to train the YOLOv8 model. When the set number of training epochs is reached, the trained YOLOv8 model is obtained.

3. The intelligent accelerated breeding cabin according to claim 1, characterized in that: The detection module includes a temperature sensor, a humidity sensor, a CO2 concentration sensor, a photoelectric sensor, and a light intensity sensor, all of which are connected to the controller respectively. The regulation module includes a refrigerator, a heater, a dehumidifier, a spray valve, a fan, and a motor, all of which are connected to the controller respectively. The detection module collects the current temperature, humidity, CO2 concentration, and light intensity and transmits them to the controller. The controller calculates the control signal based on the difference between the current temperature, humidity, CO2 concentration, and light intensity and the target parameters, and sends it to the regulation module to control the start and stop of the refrigerator and the heater, the start and stop of the dehumidifier and the spray valve, the start and stop of the solenoid valve and the fan, and the lifting of the motor, so as to adjust the temperature, humidity, CO2 concentration, and light intensity in the cabin.

4. The intelligent accelerated breeding cabin according to claim 3, wherein: The controller is a PLC with a built-in PID regulator. The parameters of the PID regulator are adjusted using an improved particle swarm optimization algorithm, and the ITAE evaluation index is used as the fitness function f i : Among them, e(t) is the system error, t is the system operation time, and the calculation formulas for the learning factors c1 and c2 are: Among them, c 1_max , c 1_min , c 2_max , c n2_min are the upper and lower limits of the learning factors c1 and c2 respectively.

5. The intelligent accelerated breeding cabin according to claim 4, characterized in that: The adaptive inertia weight factor ω(t) of the improved particle swarm optimization algorithm is: Among them, τ is the weight decay adjustment factor.

6. The intelligent accelerated breeding cabin according to claim 3, characterized in that: The temperature sensor, humidity sensor, and CO2 concentration sensor are respectively located on the side of the tray rack. The photoelectric sensor and the light intensity sensor are respectively located above the tray rack. The refrigerator is located in the equipment room. The heater, dehumidifier, and fan are respectively located on both side walls of the breeding cabin. The two side walls of the breeding cabin are respectively provided with a hot air outlet and a refrigerator air outlet. A fresh air outlet is provided on one side wall of the breeding cabin. A circulating air duct is provided in the crop cultivation area, and the circulating air duct is connected to the CO2 outlet on the top wall of the breeding cabin. A humidification pipeline connected to the spray valve is provided near the tray rack.

7. The intelligent accelerated breeding cabin according to claim 1, characterized in that: Above the tray rack, there is a multi-spectral light source installed on a lifting lamp rack, and the lifting lamp rack is fixedly installed on the top wall of the breeding cabin.

8. The intelligent accelerated breeding cabin according to claim 1, characterized in that: The breeding cabin further includes a cultivation rack located on the tray rack, a cultivation tank located on the cultivation rack, and a water and fertilizer circulation system connected to the cultivation tank. The water and fertilizer circulation system includes a water supply pipeline. The controller controls the solenoid valve and the water pump on the water supply pipeline to inject nutrient solution into the cultivation tank.

9. The intelligent accelerated breeding cabin according to claim 1, wherein: The breeding cabin further includes a cabin body and an intelligent electric door installed in the crop cultivation area, and a single-opening door is installed in the equipment area.

10. A method for regulating the cabin environment, characterized in that: Applied to the intelligent accelerated breeding cabin according to any one of claims 1-9, the method includes: Collecting the crop growth images in the cabin, preprocessing the images to obtain the preprocessed images; Inputting the preprocessed images into the visual recognition module to predict the crop growth stage and transmitting it to the host computer; The host computer generates control instructions according to the target parameters corresponding to the crop growth stage and transmits them to the controller; The detection module detects the environmental parameters in the cabin in real time and feeds them back to the controller, and the controller drives the regulation module to adjust the environmental parameters in the cabin until the target parameters are reached.

Images