Flue-cured tobacco seedling environment intelligent monitoring method based on Internet of Things

Abstract

The invention discloses a flue-cured tobacco seedling environment intelligent monitoring method based on the Internet of Things, and particularly relates to the technical field of agricultural Internet of Things, and the method comprises the steps: firstly collecting environment data through various sensor nodes, and obtaining a flue-cured tobacco seedling comprehensive environment state vector; based on a flue-cured tobacco seedling stage identification model, identifying the growth stage of flue-cured tobacco seedlings and corresponding target environment data, calculating a flue-cured tobacco seedling growth environment deviation index, triggering an environment intelligent control mode according to a deviation index abnormity analysis result, and performing environment intelligent monitoring feedback; and finally, optimizing parameters in the environment intelligent control mode by using a gradient descent algorithm based on environment parameter intelligent control process data. Based on dynamic target setting and weight adjustment in the growth stage, environment control better meets the actual physiological needs of the flue-cured tobacco seedlings, the system can automatically calculate the optimal adjusting amount according to the deviation degree and the growth stage of the tobacco seedlings, and then more stable environment conditions are provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural Internet of Things, in particular to a flue-cured tobacco seedling environment intelligent monitoring method based on Internet of Things. BACKGROUND

[0002] Flue-cured tobacco is an important crop, and its planting and management are crucial to the improvement of tobacco yield and quality. As an important link in the planting process, flue-cured tobacco seedling management directly affects the growth and yield of tobacco.

[0003] Traditional flue-cured tobacco seedling environment monitoring relies on manual patrol recording and experience-based adjustment methods, and through rough control of a single environmental factor (such as temperature), the following defects exist: only a single indicator such as temperature or humidity is focused on, and the coupling relationship between temperature, humidity, light, CO2 concentration and other environmental factors is ignored; manual inspection and manual adjustment cannot achieve real-time and accurate control, and cannot respond in time when the environment changes suddenly, resulting in large fluctuations in the growth environment of seedlings; different growth stages (such as the seedling stage, the cross stage, the rooting stage, and the seedling stage) of flue-cured tobacco seedlings have different environmental requirements, and the traditional method lacks a dynamic optimization strategy based on the growth stage; the start and stop of devices (such as heating, light supplementing, and fans) lack coordinated optimization, often causing energy waste.

[0004] Although some existing Internet of Things monitoring systems achieve remote monitoring and automatic control, their control logic is mostly simple threshold judgment, lacking deep mining and analysis ability of data. When multiple environmental parameters appear abnormal at the same time, the interaction and influence between parameters cannot be considered comprehensively, and a scientific and reasonable regulation strategy cannot be developed, thus the problem of multi-factor coordination and dynamic optimization cannot be fundamentally solved, so the control effect and intelligent degree are limited. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a flue-cured tobacco seedling environment intelligent monitoring method based on Internet of Things to solve the problems raised in the background art.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a flue-cured tobacco seedling environment intelligent monitoring method based on Internet of Things, comprising:

[0007] S1: collecting environmental data through various sensor nodes distributed in the flue-cured tobacco seedling environment, and fusing the environmental data to obtain a comprehensive environmental state vector of the flue-cured tobacco seedling;

[0008] S2: identifying the growth stage of the flue-cured tobacco seedling based on a flue-cured tobacco seedling stage identification model, and obtaining target environmental data corresponding to the current growth stage;

[0009] S3: Based on the comprehensive environmental state vector of flue-cured tobacco seedlings at the current growth stage and the corresponding target environmental data, calculate the deviation index of the growth environment of flue-cured tobacco seedlings at the current growth stage.

[0010] S4: Analyze the deviation index of the current growth stage of flue-cured tobacco seedlings, trigger the intelligent environmental control mode based on the abnormal analysis results, and generate intelligent control instructions for environmental parameters;

[0011] S5: Compare the results of the execution of the intelligent control command for environmental parameters with the target environmental data, and feed back the abnormal comparison results to S3 and S4;

[0012] S6: Based on the intelligent control process data of environmental parameters, the gradient descent algorithm is used to optimize the parameters in the intelligent environmental control mode, and the optimization results are fed back to S4.

[0013] The technical effects and advantages of this invention are as follows:

[0014] 1. This invention achieves comprehensive, real-time, and accurate monitoring of environmental parameters such as temperature, humidity, light, and CO2 concentration by rationally arranging multiple sensor nodes in the tobacco seedling shed; at the same time, through data preprocessing and analysis, it effectively eliminates data noise and errors, providing a reliable data foundation for subsequent environmental control and evaluation of deviations in the tobacco seedling growth environment;

[0015] 2. Based on the dynamic target setting and weight adjustment of the growth stage, this invention makes the environmental control more in line with the actual physiological needs of flue-cured tobacco seedlings. The system can automatically calculate the optimal adjustment amount according to the degree of deviation and the growth stage of the tobacco seedlings. Through intelligent control priority sorting, it achieves coordinated control and provides more suitable and stable environmental conditions for the growth of flue-cured tobacco seedlings.

[0016] 3. This invention utilizes self-learning capabilities to deeply mine and analyze data from the intelligent control process of environmental parameters, continuously optimizing control parameters. This improves the system's control accuracy and adaptability over time, further enhancing management efficiency and seedling quality, reducing production costs, and increasing the economic benefits of flue-cured tobacco cultivation. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall process of the present invention.

[0018] Figure 2 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, the present invention provides an intelligent monitoring system for flue-cured tobacco seedling environment based on the Internet of Things, including an environmental data perception and fusion module, a growth stage management and target setting module, an intelligent monitoring and evaluation module for the growth environment, an intelligent decision-making and control module, an intelligent environmental monitoring feedback module, and an intelligent environmental monitoring self-learning optimization module.

[0021] The environmental data perception and fusion module is connected to the growth stage management and target setting module. The intelligent monitoring and evaluation module of the growth environment is connected to both the growth stage management and target setting module and the intelligent decision-making and control module. The intelligent environmental monitoring feedback module is connected to both the intelligent decision-making and control module and the intelligent environmental monitoring self-learning optimization module. The intelligent environmental monitoring self-learning optimization module is connected to both the intelligent monitoring and evaluation module of the growth environment and the intelligent decision-making and control module.

[0022] Environmental data perception and fusion module: Collects environmental data through various sensor nodes distributed in the flue-cured tobacco seedling environment, fuses the environmental data to obtain a comprehensive environmental state vector of flue-cured tobacco seedling, and transmits it to the growth stage management and target setting module;

[0023] Growth stage management and target setting module: Based on the tobacco seedling stage identification model, it identifies the growth stage of tobacco seedlings, obtains the target environment data corresponding to the current growth stage, and transmits it to the intelligent decision and control module;

[0024] Intelligent monitoring and evaluation module for growth environment: Based on the comprehensive environmental state vector of flue-cured tobacco seedlings at the current growth stage and the corresponding target environmental data, calculate the deviation index of the growth environment of flue-cured tobacco seedlings at the current growth stage and transmit it to the intelligent decision-making and control module.

[0025] Intelligent Decision and Control Module: Analyzes the deviation index of the flue-cured tobacco seedling growth environment at the current growth stage, triggers the intelligent environmental control mode based on the anomaly analysis results, and generates intelligent control instructions for environmental parameters;

[0026] Environmental intelligent monitoring and feedback module: compares the results of the execution of intelligent control commands for environmental parameters with the target environmental data, and feeds back the abnormal comparison results to the intelligent decision-making and control module;

[0027] Environmental intelligent monitoring self-learning optimization module: Based on the environmental parameter intelligent control process data, it uses the gradient descent algorithm to optimize the parameters in the environmental intelligent control mode and feeds the optimization results back to the intelligent decision and control module;

[0028] Please see Figure 2 As shown, the intelligent monitoring method for flue-cured tobacco seedling environment based on the Internet of Things includes: S1: Collecting environmental data through various sensor nodes distributed in the flue-cured tobacco seedling environment, and fusing the environmental data to obtain a comprehensive environmental state vector of flue-cured tobacco seedlings; S2: Identifying the growth stage of flue-cured tobacco seedlings based on the flue-cured tobacco seedling stage identification model, and obtaining the target environmental data corresponding to the current growth stage; S3: Calculating the deviation index of the flue-cured tobacco seedling growth environment at the current growth stage based on the comprehensive environmental state vector of the flue-cured tobacco seedlings at the current growth stage and the corresponding target environmental data; S4: Analyzing the deviation index of the flue-cured tobacco seedling growth environment at the current growth stage, triggering the intelligent environmental control mode based on the anomaly analysis results, and generating intelligent control commands for environmental parameters; S5: Comparing the results after the execution of the intelligent environmental parameter control commands with the target environmental data, and feeding back the anomaly comparison results to S4; S6: Optimizing the parameters in the intelligent environmental control mode using the gradient descent algorithm based on the intelligent environmental parameter control process data, and feeding back the optimization results to S4.

[0029] S1: Environmental data is collected by various sensor nodes distributed in the flue-cured tobacco seedling environment, and the environmental data is fused to obtain the comprehensive environmental state vector of flue-cured tobacco seedling, including the following steps:

[0030] S1.1: The collection period is set to T by various sensor nodes distributed in the flue-cured tobacco seedling environment. In the i-th period, the first environmental data is collected, including temperature, humidity, light intensity, CO2 concentration and soil moisture.

[0031] S1.2: Preprocess the collected environmental data, including data cleaning, data denoising and data alignment, to obtain the second environmental data, including preprocessed temperature, humidity, light intensity, CO2 concentration and soil moisture;

[0032] In this embodiment, it is necessary to specifically explain that data cleaning refers to removing invalid values, illogical values, and other abnormal values; data denoising refers to smoothing using median filtering or mean filtering; data alignment uses the system's unified clock as a reference to align the timestamps of all sensor data, and for gaps caused by removing invalid data, forward padding or linear interpolation is used to repair the data.

[0033] S1.3: Take the average value of the second environmental data collected by all sensors of the same type and fuse them to obtain the comprehensive environmental state vector E(i) of flue-cured tobacco seedling cultivation, E(i)=[T(i),H(i),L(i),C(i),S(i)], where T(i), H(i), L(i), C(i) and S(i) are the average values ​​of environmental factors such as temperature, humidity, light intensity, CO2 concentration and soil moisture in the second environmental data in the i-th period, respectively.

[0034] S2: Based on the tobacco seedling stage identification model, identify the growth stage of tobacco seedlings and obtain the target environment data corresponding to the current growth stage, including the following steps:

[0035] S2.1: Constructing a model for identifying the seedling stage of flue-cured tobacco:

[0036] S2.1.1: Constructing a flue-cured tobacco seedling calendar: First, predefine the growth stage sequence G of flue-cured tobacco seedlings within the system, including the emergence stage G1, the cross-shaped stage G2, the rooting stage G3, and the seedling maturity stage G4; then, set the baseline duration range [t_min(G), t_max(G)] for each stage, where t_min(G) and t_max(G) are the shortest and longest values ​​of the baseline duration of growth stage G, respectively. For example, stage G1 may last [10, 14] days under normal conditions; finally, construct a flue-cured tobacco seedling calendar for each growth stage.

[0037] In this embodiment, it is necessary to specifically explain that the emergence period refers to the period from sowing to emergence, that is, from sowing to 50% of the seedlings emerging from the soil; the cross-shaped period refers to the period from emergence to the cross-shaped period, that is, from 50% of the seedlings emerging from the soil to 50% of the seedlings showing their first pair of true leaves; the rooting period refers to the period from the cross-shaped period to the rooting period, that is, from the first pair of true leaves to the critical period of seedling root formation; and the seedling maturity period refers to the period from the rooting period to the seedling maturity, that is, from the formation of the root system to reaching the transplanting standard.

[0038] S2.1.2: Constructing the accumulated temperature model: First, the system is set to operate on a daily basis, with each day divided into n time intervals, and the duration of each time interval is Δt. j Calculate the effective accumulated temperature DD_d for the day. T j,avg T and T0 are the average temperature and the baseline temperature for tobacco seedling growth, respectively, for the j-th time interval (e.g., 1 hour). j,avg When T < T0, the contribution of this item is 0. Usually, T0 = 10℃ is set. When the temperature is below 10℃, the seedling basically stops growing and does not accumulate effective heat. Then, starting from the beginning of each growth stage, the effective temperature DD_d is accumulated day by day to obtain the accumulated temperature T(DD_d) of the current growth stage. T(DD_d) = ∑DD_d.

[0039] S2.1.3: Growth Stage Identification and Switching Judgment: First, starting from the sowing date (or the start date set by the administrator), the system state is initialized to G=G1, and T(DD_d) is cleared to zero; then, in each effective accumulated temperature calculation cycle (e.g., at dawn every day), the current growth stage G is read. cu Calculate yesterday's effective accumulated temperature DD_d, and update the current stage's accumulated temperature T(DD_d) = T(DD_d) + DD_d; if T(DD_d) ≥ the current growth stage's accumulated temperature threshold T(DD_d) G,th The number of days is greater than or equal to the threshold d th If (e.g., 2 days) the current stage is considered complete, the system will switch the state to the next growth stage G. next And reset the accumulated temperature T(DD_d). next =0; if the current date has exceeded the current growth stage G cu The latest possible end date is the sowing date + t_max(G), but if T(DD_d) is less than the threshold, the system forcibly switches to the next growth stage G. next To prevent seedlings from becoming "stunted" due to prolonged stagnation at a certain stage, and to ensure the smooth progress of seedling cultivation; if the current date has not exceeded the current growth stage G... cu The earliest possible end date is the sowing date + t_min(G) and T(DD_d) ≥ the cumulative accumulated temperature threshold T(DD_d) for the current growth stage. G,th The growth stage does not switch immediately, but waits until the earliest end date and T(DD_d) ≥ the cumulative accumulated temperature threshold T(DD_d) of the current growth stage. G,th Switch again to prevent misjudgment of the stage due to short-term abnormal high temperature;

[0040] This embodiment specifically explains how the fusion strategy of "defining large stages with a baseline calendar and driving small stage switching with dynamic accumulated temperature" enables automatic identification of the growth stage G; the system presets a stage accumulated temperature threshold T (DD_d) for each growth stage. G,th For example, the accumulated temperature T(DD_d) required for G1. G1,th =120℃, the accumulated temperature T(DD_d) required for G2 G2,th =180℃, and so on.

[0041] S2.3: Based on the tobacco seedling stage identification model, identify the current growth stage G of the tobacco seedlings. cu And based on historical data, set corresponding target environment data for this stage, including target environment vector E. tar (G cu ) and allowable fluctuation range [E _min (G cu ),E _mmax (G cu )], E tar(G cu )=[T(G cu ),H(G cu ),L(G cu ),C(G cu ),S(G cu )],T(G cu ), H(G cu ), L(G cu ), C(G cu ) and S(G cu These represent the target temperature, target humidity, target light intensity, target CO2 concentration, and target soil moisture, respectively. For example, during the seedling stage (G1): T(G1) = 28℃, H(G1) = 85%; during the mature seedling stage (G4): T(G4) = 22℃, H(G4) = 60%.

[0042] This embodiment specifically explains the historical data stored in the system, including the environmental state vector E(k) and the corresponding seedling growth performance at that stage (growth indicators obtained through manual input or image analysis, such as emergence rate, number of leaves, stem height, stem diameter, root system score, etc.). Then, statistical methods (such as regression analysis and correlation analysis) are used to find the correlation between historical environmental data and final seedling growth indicators. Next, the batches with the best growth performance (such as thick stems and well-developed root systems) are selected, and the environmental conditions experienced by these batches at each growth stage are analyzed retrospectively to obtain the target environmental vector E for that stage. tar (G cu ) and allowable fluctuation range [E _min (G cu ),E _mmax (G cu Finally, calculate the mode or cluster center of the environmental data of these "top student" batches at each stage, and use it as the new optimized target value E_target_optimized(G); if historical data shows that, locally, when the average temperature of the cross-stage G2 is stable at 20℃, the seedling stems are thicker, then the system can fine-tune the temperature target of T_t(G2) to 20℃.

[0043] S3: Based on the comprehensive environmental state vector of flue-cured tobacco seedlings at the current growth stage and the corresponding target environmental data, calculate the growth environment deviation index (DKI) of flue-cured tobacco seedlings at the current growth stage. ,5 represents the five parameters in the integrated environment state vector, w L Let ηE(i) be the weight of the Lth environmental factor. L Let L be the deviation of the Lth environmental factor within the i-th period. E L E tar,L and (E) - max-E- min) L These represent the value of the Lth environmental factor, the target value of the Lth environmental factor, and the difference between the maximum and minimum allowable values ​​of the Lth environmental factor, respectively.

[0044] S4: Analyze the deviation index of the flue-cured tobacco seedling growth environment at the current growth stage, trigger the intelligent environmental control mode based on the anomaly analysis results, and generate intelligent control instructions for environmental parameters, including the following steps:

[0045] S4.1: Analyze the deviation index (DKI) of the growth environment for flue-cured tobacco seedlings at the current growth stage. If DKI ≤ the corresponding threshold DKI... th Then, it continues to determine whether each parameter falls within the corresponding allowable fluctuation range. If all parameters fall within this range, the environmental intelligent monitoring is considered to be working well. Otherwise, if any parameter does not fall within this range, the target parameter environmental intelligent control mode ΔE is triggered. adj ΔE adj,L =k L ×(|E L -E tar,L |×w L ), ΔE adj,L Let k be the control quantity for the Lth environmental factor. L E L and E tar,L Let w be the adjustment coefficient, the value of the Lth environmental factor, and the target value of the Lth environmental factor, respectively. L The weight of the Lth environmental factor (adjustment coefficient k) L Sensitivity and weights used to control the adjustment L (This reflects the importance of the environmental factor), if E L and E tar,L If the difference is greater than 0, then an intelligent control command to reduce the target parameter is generated; if E L and E tar,L If the difference is less than 0, an intelligent control command to increase the target parameter is generated. The adjustment coefficient of the Lth environmental factor can be obtained by fitting historical data. For example, when the system determines that the temperature needs to be lowered, the fan is started to increase the ventilation volume in order to remove heat and lower the temperature; when the humidity is insufficient, the humidifier is started to increase the air humidity.

[0046] This embodiment requires specific explanation regarding DKI ≤ the corresponding threshold DKI. thWhen the temperature control actuator is not limited to fans and heaters, the humidity control actuator is not limited to humidifiers, the light intensity control actuator is not limited to supplemental lighting, the CO2 concentration control actuator is not limited to CO2 generators, and the soil moisture control actuator is not limited to solenoid valves or electric valves. The switch that controls the "on / off" of the irrigation water circuit receives instructions from the control module and automatically turns it on or off, thereby controlling whether to supply water to the seedbed.

[0047] S4.2: If DKI > the corresponding threshold DKI th When an abnormality is detected in the intelligent environmental monitoring system, the environmental parameter collaborative control mode is triggered, and the environmental parameter control quantity dataset D(ΔE) is obtained. adj ), D(ΔE adj )=[ΔE adj,1 ,ΔE adj,1 ,...,ΔE adj,L ], ΔE adj,L Let be the control variable for the Lth environmental factor; then, based on the deviation ηE(i) of each parameter... L The magnitude of environmental factors is intelligently prioritized for control, and the adjustment amount of low-priority environmental factors is corrected through a compensation mode ΔE. ΔE L ΔE represents the change in low-priority environmental factors during the control of high-priority environmental factors. L =|E L,af -E L,be | / E L,be E L,af and E L,be These represent the values ​​of the low-priority environmental factor after control and the value before control, respectively, during the control process of the high-priority environmental factor. If ΔE < 0, it is considered as 0. Similarly, according to E... L and E tar,L The difference generates intelligent control commands that reduce or increase the target parameters. For example, in a high temperature and high humidity scenario: {start the fan, turn off the humidifier}, the priority is ventilation over cooling. In a low temperature and low light scenario: {start the heater, turn on the supplementary light}, the priority is heat preservation over supplementary lighting.

[0048] In this embodiment, it is necessary to specifically explain that when there is a conflict in the environmental parameter control scenario, such as the temperature needs to be increased by ΔT=+2℃, but heating will reduce the humidity by 3%, and the current humidity is already 2% lower than the target value. In this case, an additional 3% humidity needs to be added, so the final humidity adjustment amount is 2%+3%.

[0049] S5: Compare the results of executing the intelligent control command for environmental parameters with the target environmental data, feed back any abnormal comparison results to S4, and calculate the deviation ηE(i) of the Lth environmental factor after execution. L If the deviation exceeds the corresponding allowable range, the target parameter intelligent control command will be readjusted again until the deviation falls within the corresponding allowable range; otherwise, no further adjustment is needed.

[0050] S6: Based on the intelligent control process data of environmental parameters, the parameters in the intelligent environmental control mode are optimized using the gradient descent algorithm, and the optimization results are fed back to S3 and S4, including the following steps:

[0051] S6.1: First, based on N sets of environmental parameter intelligent control process data, including the comprehensive environmental state vector E(i) collected at the beginning of the i-th cycle and the environmental parameter control quantity ΔE adj (i) the growth stage G(i) and the environmental state vector E(i+1) collected at the beginning of the (i+1)th cycle, i.e., the execution of ΔE adj (i) The result is then used; then, using the gradient descent algorithm, the weight and adjustment coefficient k of the Lth environmental factor in the intelligent environmental control mode are adjusted through the loss function J(w,k). L Optimize, Let I represent the I-th set of data, and w and k represent the set of parameters to be optimized, i.e., all weights w L and all adjustment coefficients k L λ is the regularization coefficient (used to balance "control precision" and "control cost"), Ω(ΔE) adj (i) is the regularization term, which is the sum of squares of the ratios of the output of all actuators in the control process based on the control quantity to the maximum rated output (e.g., fan speed, heater power, valve opening, etc.).

[0052] S6.2: Minimize the loss function J(w,k) using the gradient descent algorithm to obtain the parameter update formula θ. new θ new =θ old -η×∇ θ J(θ), where θ is the set of parameters for w and k to be optimized. new and θ old Here, η represents the parameters before and after the update, respectively, and η is the learning rate, controlling the step size for each parameter update. θ J(θ) is the gradient of the loss function J(w,k) with respect to the parameter θ;

[0053] S6.3: When the difference between the loss function J(w,k) before and after the iteration is less than the corresponding convergence threshold for m consecutive times (e.g., 5 consecutive times) or the maximum number of iterations is reached, it indicates that the machine learning algorithm has converged. The current optimal parameters w and k are output, and the weights are fed back to S3 and S4, and the adjustment coefficient is fed back to S4.

[0054] This embodiment specifically explains that, through the action of regularization terms, the system aims to minimize the difference between the result after execution control and the target result, while also selecting a mild and energy-efficient control strategy.

[0055] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0056] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring of tobacco seedling raising environment based on the Internet of Things, characterized in that: include: S1: Collect environmental data by various sensor nodes distributed in the flue-cured tobacco seedling environment, and fuse the environmental data to obtain the comprehensive environmental state vector of flue-cured tobacco seedling. S2: Based on the tobacco seedling stage identification model, identify the growth stage of tobacco seedlings and obtain the target environment data corresponding to the current growth stage; S3: Based on the comprehensive environmental state vector of flue-cured tobacco seedlings at the current growth stage and the corresponding target environmental data, calculate the deviation index of the growth environment of flue-cured tobacco seedlings at the current growth stage. S4: Analyze the deviation index of the current growth stage of flue-cured tobacco seedlings, trigger the intelligent environmental control mode based on the abnormal analysis results, and generate intelligent control instructions for environmental parameters; S5: Compare the results of the intelligent control command for environmental parameters with the target environmental data, and feed back any abnormal comparison results to S4; S6: Based on the intelligent control process data of environmental parameters, the gradient descent algorithm is used to optimize the parameters in the intelligent environmental control mode, and the optimization results are fed back to S3 and S4.

2. The intelligent monitoring method for flue-cured tobacco seedling raising environment based on the Internet of Things according to claim 1, characterized in that: The tobacco seedling stage identification model in S2 includes: S2.1.1: Constructing a flue-cured tobacco seedling calendar: First, predefine the growth stage sequence G of flue-cured tobacco seedlings within the system, including the emergence stage G1, the cross-shaped stage G2, the rooting stage G3, and the seedling maturity stage G4; then, set the baseline duration range [t_min(G), t_max(G)] for each stage, where t_min(G) and t_max(G) are the shortest and longest values ​​of the baseline duration of the growth stage G, respectively; finally, construct a flue-cured tobacco seedling calendar for each growth stage. S2.1.2: Constructing the accumulated temperature model: First, the system is set to operate on a daily basis, with each day divided into n time intervals, and the duration of each time interval is Δt. j Calculate the effective accumulated temperature DD_d for the day. T j,avg T and T0 are the average temperature and the baseline temperature for tobacco seedling growth in the j-th time interval, respectively. When T j,avg When <T0, the contribution of this term is 0; then starting from the beginning of each growth stage, the effective temperature DD_d is accumulated daily to obtain the accumulated temperature T(DD_d) of the current growth stage, T(DD_d)=∑DD_d.

3. The intelligent monitoring method for flue-cured tobacco seedling raising environment based on the Internet of Things according to claim 1, characterized in that: The tobacco seedling stage identification model in S2 also includes: S2.1.3: Growth Stage Identification and Switching Judgment: First, starting from the sowing date, the system state is initialized to G=G1, and T(DD_d) is cleared to zero; then, in each effective accumulated temperature calculation cycle, the current growth stage G is read. cu Calculate yesterday's effective accumulated temperature DD_d, and update the current stage's accumulated temperature T(DD_d) = T(DD_d) + DD_d; if T(DD_d) ≥ the current growth stage's accumulated temperature threshold T(DD_d) G,th The number of days is greater than or equal to the threshold d th If the current stage is determined to be complete, the system will switch the state to the next growth stage G. next And reset the accumulated temperature T(DD_d). next =0; if the current date has exceeded the current growth stage G cu The latest possible end date is the sowing date + t_max(G), but if T(DD_d) is less than the threshold, the system forcibly switches to the next growth stage G. next If the current date has not exceeded the current growth stage G cu The earliest possible end date is the sowing date + t_min(G) and T(DD_d) ≥ the cumulative accumulated temperature threshold T(DD_d) for the current growth stage. G,th The growth stage does not switch immediately, but waits until the earliest end date and T(DD_d) ≥ the cumulative accumulated temperature threshold T(DD_d) of the current growth stage. G,th Switch again.

4. The intelligent monitoring method for flue-cured tobacco seedling environment based on the Internet of Things according to claim 1, characterized in that: The target environment data corresponding to the current growth stage in S2: Based on the tobacco seedling stage identification model, the current growth stage G of the tobacco seedlings is identified. cu Based on historical data, corresponding target environment data is set for this stage, including the target environment vector E. tar (G cu ) and allowable fluctuation range [E _min (G cu ),E _mmax (G cu )], E tar (G cu )=[T(G cu ),H(G cu ),L(G cu ),C(G cu ),S(G cu )],T(G cu ), H(G cu ), L(G cu ), C(G cu ) and S(G cu These are the target temperature, target humidity, target light intensity, target CO2 concentration, and target soil moisture, respectively.

5. The intelligent monitoring method for flue-cured tobacco seedling raising environment based on the Internet of Things according to claim 1, characterized in that: The deviation index (DKI) of the current growth stage of flue-cured tobacco seedlings in S3. ,5 represents the five parameters in the integrated environment state vector, w L Let ηE(i) be the weight of the Lth environmental factor. L denoted as the deviation of the Lth environmental factor within the i-th period.

6. The intelligent monitoring method for flue-cured tobacco seedling environment based on the Internet of Things according to claim 1, characterized in that: The implementation of S4 includes: S4.1: Analyzing the deviation index (DKI) of the flue-cured tobacco seedling growth environment at the current growth stage; if DKI ≤ the corresponding threshold DKI... th Then, it continues to determine whether each parameter falls within the corresponding allowable fluctuation range. If all parameters fall within this range, the environmental intelligent monitoring is considered to be working properly. Otherwise, if any parameter does not fall within this range, the target parameter environmental intelligent control mode ΔE is triggered. adj ΔE adj,L =k L ×(|E L -E tar,L |×w L ), ΔE adj,L Let k be the control quantity for the Lth environmental factor. L E L and E tar,L Let w be the adjustment coefficient, the value of the Lth environmental factor, and the target value of the Lth environmental factor, respectively. L Let E be the weight of the Lth environmental factor. L and E tar,L If the difference is greater than 0, then an intelligent control command to reduce the target parameter is generated; if E L and E tar,L If the difference is less than 0, an intelligent control command is generated to increase the target parameter.

7. The intelligent monitoring method for flue-cured tobacco seedling environment based on the Internet of Things according to claim 1, characterized in that: The implementation of S4 further includes: S4.2: If DKI > the corresponding threshold DKI th When an abnormality is detected in the intelligent environmental monitoring system, the environmental parameter collaborative control mode is triggered, and the environmental parameter control quantity dataset D(ΔE) is obtained. adj ), D(ΔE adj )=[ΔE adj,1 ,ΔE adj,1 ,...,ΔE adj,L ], ΔE adj,L Let be the control variable for the Lth environmental factor; then, based on the deviation ηE(i) of each parameter... L The magnitude of the environmental factors is used to intelligently prioritize and control them, and the adjustment amount of low-priority environmental factors is corrected through a compensation mode ΔE. If ΔE < 0, it is considered as 0. Similarly, based on E... L and E tar,L The difference generates intelligent control commands that either reduce or increase the target parameters.

8. The intelligent monitoring method for flue-cured tobacco seedling environment based on the Internet of Things according to claim 1, characterized in that: The anomaly comparison results in S5: Calculate the deviation ηE(i) of the Lth environmental factor after execution. L If the deviation exceeds the corresponding allowable range, the target parameter intelligent control command will be readjusted again until the deviation falls within the corresponding allowable range; otherwise, no further adjustment is needed.

Images