Smart farm environment regulation and control method and system based on big data analysis

Abstract

The invention relates to the technical field of agricultural management, in particular to a smart farm environment regulation and control method and system based on big data analysis, and the method comprises the following steps: obtaining an instantaneous temperature value of the upper surface of a leaf and an instantaneous temperature value of the lower surface of the leaf through an infrared temperature sensor, obtaining an instantaneous value of stomatal conductance through a stomatal conductance sensor; and acquiring a water vapor pressure difference value of the air cavity through a water vapor pressure sensor, aligning a time sequence by adopting a multi-source data fusion algorithm, eliminating noise, and generating a dynamic physiological data set. According to the method, the leaf temperature difference, the air hole conductance and the air cavity water vapor pressure difference are collected, a real-time dynamic data set is constructed, a temperature difference amplification signal is extracted, transpiration enhancement triggering, wind speed increasing, water mist particle size decreasing and wind speed and particle size linkage regulation and control are analyzed in combination with the trend, a regulation and control weight is constructed by introducing multiple regression analysis, and the wind speed and the particle size are dynamically regulated according to the weight. A leaf temperature difference regression base line is guaranteed, the regulation and control accuracy and physiological adaptability are improved, and the environment self-adaptive regulation capability is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of agricultural management technology, and in particular to a smart farm environment control method and system based on big data analysis. Background Art

[0002] The field of agricultural management technology encompasses digital, refined, and intelligent management methods and systems for the entire agricultural production process, leveraging technologies such as information communication, big data analysis, and automated control. Its core areas include crop growth environment monitoring, agricultural activity scheduling, optimized agricultural resource allocation, and pest and disease early warning and control. This technology utilizes a sensory layer to collect data related to the farm environment and crop growth, then utilizes computing platforms to process and analyze this data, formulate control strategies, and transmit them to the execution end. This enables remote control of agricultural activities, intelligent resource allocation, and improved management efficiency, thereby promoting the transformation of traditional agriculture to smart agriculture.

[0003] Among them, the smart farm environmental control method refers to collecting real-time data on key environmental factors that affect crop growth, such as temperature and humidity, light, soil moisture, and carbon dioxide concentration, during the agricultural production process. By establishing a relationship model between environmental factors and crop growth status, and utilizing multi-source heterogeneous data fusion technology and rule-driven parameter setting logic, a basis for control decision-making is formed. Combined with specific execution equipment such as electric curtains, fans, and water pumps, the internal environment of agricultural facilities is automatically adjusted according to the model calculation results. This method is usually based on intelligent sensor networks, combined with decision-making models based on agricultural knowledge graphs or regression prediction mechanisms based on historical data training, to achieve dynamic management and closed-loop control of facility environmental parameters.

[0004] Current methods are primarily based on environmental parameter thresholds, fail to incorporate crop physiological status indicators, and lack a response mechanism to actual transpiration changes, leading to delayed control when crops enter stress states. The regulation process fails to consider the synergistic effect between wind speed and atomized particle size, which can easily lead to uneven cooling efficiency or waste of resources. The correlation between parameters is not modeled, and the regulation strategy is based on a single factor, ignoring the influence of water vapor coupling, resulting in a deviation between the regulation results and the actual needs of the crop. The feedback mechanism is based on static settings and does not implement dynamic regulation, resulting in a slow response and difficulty in coping with rapid changes in physiological status. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a smart farm environment control method and system based on big data analysis.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solution: a smart farm environment control method based on big data analysis, comprising the following steps:

[0007] S1: The instantaneous temperature values of the upper and lower surfaces of the leaves are obtained through infrared temperature sensors, the instantaneous stomatal conductance value is obtained through stomatal conductance sensors, and the air cavity water vapor pressure difference value is obtained through water vapor pressure sensors. A multi-source data fusion algorithm is used to align the time series and remove noise to generate a dynamic physiological data set.

[0008] S2: performing a difference calculation on the upper and lower surface temperatures of the leaf in the dynamic physiological data set, comparing the temperature difference values of adjacent cycles, and using time series trend analysis to determine the temperature difference increase. If the increase exceeds a threshold for three consecutive cycles, a transpiration activity enhancement signal is output;

[0009] S3: Based on the transpiration activity enhancement signal, the wind speed of the evaporative cooling unit is adjusted to increase by 0.5 m / s per cycle and the water mist particle size is adjusted to decrease by 5% per cycle, thereby generating a wind speed-particle size coordinated control instruction;

[0010] S4: Performing a Pearson correlation calculation on the stomatal conductance and the air cavity water vapor pressure difference in the dynamic physiological data set. If the absolute value of the correlation coefficient is greater than 0.8, a multivariate linear regression model is used to assign a weight coefficient to generate a water vapor linkage control coefficient.

[0011] As a further solution of the present invention, the dynamic physiological data set specifically includes the instantaneous temperature value of the upper surface of the leaf, the instantaneous temperature value of the lower surface of the leaf, the instantaneous value of the stomatal conductance, and the air cavity water vapor pressure difference value. The transpiration activity enhancement signal includes the temperature difference increase, the number of consecutive cycles, and the temperature difference threshold. The wind speed-particle size coordinated control instruction includes the wind speed adjustment gradient and the particle size adjustment gradient. The water vapor linkage control coefficient specifically refers to the correlation coefficient threshold and the regression weight.

[0012] As a further solution of the present invention, the multi-source data fusion algorithm includes a timestamp alignment algorithm, a sliding window standard deviation calculation method, a wavelet threshold denoising method and a weighted Kalman filtering algorithm;

[0013] The wind speed adjustment gradient and the particle size adjustment gradient are parameter solutions after optimizing the transpiration suppression efficiency based on a multiple regression model;

[0014] The determination threshold of the absolute value of the correlation coefficient is the critical value verified by the t-test under the conditions of significance level α=0.01 and degree of freedom n-2;

[0015] The feedback control algorithm is a dynamic parameter mapping rule based on proportional-integral-differential, with a proportional coefficient set to 0.8, an integral time constant to 10 seconds, and a differential time constant to 2 seconds.

[0016] As a further solution of the present invention, the steps of acquiring the dynamic physiological data set are specifically as follows:

[0017] S101: Collect the instantaneous values of the upper and lower surface temperatures of the blade output by the infrared temperature sensor, synchronously obtain the instantaneous conductance value output by the stomatal conductance sensor and the instantaneous value of the water vapor pressure difference output by the water vapor pressure sensor, and use a timestamp alignment algorithm to match the four types of parameters according to millisecond-level timestamps to generate synchronized time series parameters;

[0018] The timestamp alignment algorithm uses the Coordinated Universal Time (UTC) time base to uniformly convert the timestamps of the sensor acquisition parameters into millisecond-level UTC time;

[0019] S102: Based on the synchronization time series parameters, a sliding window standard deviation calculation method is used to remove outliers from the temperature value sequence, and a wavelet threshold denoising is performed on the stomatal conductance value sequence and the water vapor pressure difference value sequence, and the low-frequency features are retained after the high-frequency noise components are truncated to generate a reconstructed feature sequence;

[0020] The wavelet threshold denoising adopts the Stein unbiased risk estimation method to determine the threshold and cut off the noise component with a frequency higher than 1kHz;

[0021] S103: calling the reconstructed feature sequence, calculating the instantaneous value of the temperature difference between the upper and lower surfaces of the leaf, combining the instantaneous value of the stomatal conductance and the instantaneous value of the water vapor pressure difference, dynamically fusing the multi-dimensional parameters using a weighted Kalman filter algorithm, and assigning fusion weights based on the covariance matrix to generate a dynamic physiological data set;

[0022] The fusion weights of the weighted Kalman filter algorithm are allocated according to the normalized values of the diagonal elements of the covariance matrix.

[0023] As a further solution of the present invention, the step of obtaining the transpiration activity enhancement signal is specifically as follows:

[0024] S201: Obtaining upper and lower surface temperature data of each leaf in the dynamic physiological data set, aligning the upper and lower surface temperatures of the same leaf at the same time along the time axis using the sampling time as an index, performing a subtraction operation at each time point, calculating the temperature difference between the two surfaces of the leaf per unit time, and generating a temperature difference sequence corresponding to multiple time points;

[0025] The unit time is the sensor sampling interval, which is set to 100 milliseconds;

[0026] S202: Based on the temperature difference sequence, data segments of two adjacent cycles are intercepted according to the period division rule. For the temperature difference at each time point, the operation of subtracting the value of the previous cycle from the value of the next cycle is performed. The temperature difference changes at the corresponding time points between the cycles are calculated. All changes are arranged in chronological order to generate a temperature difference increase sequence for adjacent cycles.

[0027] The cycle division rule is to define every 300 consecutive sampling points as a control cycle;

[0028] S203: Call the adjacent period temperature difference amplification sequence, take three consecutive periods as the window unit, perform a summation operation on the amplification values of multiple time points in the window, compare the summation result with the preset transpiration activity amplification threshold item by item, and generate a transpiration activity enhancement signal when the cumulative amplification of any window exceeds the threshold.

[0029] The transpiration activity increase threshold is a temperature difference cumulative increase range of 3.2°C to 4.5°C calibrated by the gradient descent method optimization experiment.

[0030] As a further solution of the present invention, the steps for obtaining the wind speed-particle size coordinated control instruction are specifically as follows:

[0031] S301: Based on the transpiration activity enhancement signal, the current wind speed value is called, and an incremental step is superimposed on the current value. The initial wind speed and a fixed step parameter are input into a periodic increasing function, and the superimposed value of the wind speed in each period is calculated to generate a wind speed increment value.

[0032] The incremental step parameter is calibrated by the fluid mechanics model to calibrate the wind duct resistance coefficient of the evaporative cooling unit, and the step unit is m / s. 2 ;

[0033] S302: Dividing nodes according to the period of the wind speed increment value, extracting the median value of the original water mist particle size distribution, reducing the upper and lower limits of the particle size range simultaneously in a geometrically decreasing manner, and compressing and translating the particle size distribution curve using a segmented iterative algorithm to generate a particle size decreasing gradient;

[0034] The compression translation ratio of the segmented iterative algorithm is to reduce the particle size distribution interval width by 5% per cycle;

[0035] S303: On the control cycle time axis, a Cartesian product mapping is performed on the numerical sequence of the wind speed increment value and the distribution parameter of the particle size decrement gradient to establish a two-dimensional parameter combination matrix of wind speed and particle size, and an instruction parameter set is generated through matrix coordinate transformation to obtain a wind speed and particle size coordinated control instruction;

[0036] The matrix coordinate transformation rule is that the wind speed-particle size coupling coefficient is 0.7, and the linear correlation between wind speed and particle size is calibrated through experiments.

[0037] As a further solution of the present invention, the step of obtaining the water vapor linkage control coefficient is specifically as follows:

[0038] S401: detecting the stomatal conductance and the air cavity water vapor pressure difference in the dynamic physiological data set, and calculating the covariance and standard deviation ratio between the stomatal conductance and the air cavity water vapor pressure difference based on the time series change value of the stomatal conductance and the time series change value of the air cavity water vapor pressure difference, to generate a Pearson correlation coefficient value;

[0039] S402: calling the Pearson correlation coefficient value, comparing the absolute value of the correlation coefficient with a preset correlation threshold, and if the condition is met, establishing a linear combination equation for the stomatal conductance and the air cavity water vapor pressure difference by the least squares method to generate a weight distribution reference parameter;

[0040] S403: Calling the weight distribution benchmark parameter, extracting the difference between the stomatal conductance at adjacent time points and the difference between the air cavity water vapor pressure difference at adjacent time points, integrating the ambient temperature sensor acquisition value and the leaf surface area measurement value, and using the formula:

[0041]

[0042] Calculate the ratio of the square root of the dynamic change difference to the weighted sum of the ambient temperature, and combine it with the regression model residual correction to generate the water vapor linkage control coefficient;

[0043] Where W represents the water vapor linkage control coefficient, ΔG s is the difference in stomatal conductance at adjacent time points, ΔV p is the difference in water vapor pressure difference between adjacent time points in the air cavity, λ i is the ambient temperature sensor calibration coefficient, is the real-time measurement value of the temperature sensor, n is the number of sensors, r is the Pearson correlation coefficient, S leaf is the calibrated value of the blade surface area;

[0044] The ambient temperature sensor calibration coefficient λ i The leaf surface area calibration value S is obtained by least square fitting the nonlinear relationship between ambient temperature and stomatal conductance. leaf is the mean value of the measured leaf projected area.

[0045] As a further embodiment of the present invention, the method further comprises:

[0046] S5: Integrate the wind speed-particle size coordinated control instruction and the water vapor linkage control coefficient, and dynamically map them to the evaporative cooling unit using a feedback control algorithm. When the water vapor linkage control coefficient is greater than 0.6, the wind speed is increased to 1.2 times the current value, and the water mist particle size is reduced to 90%. Continue to adjust until the blade surface temperature difference returns to the set baseline;

[0047] The control parameters dynamically mapped by the feedback control algorithm include wind speed multiples and particle size percentages.

[0048] As a further solution of the present invention, the step S5 is specifically as follows:

[0049] S501: Integrate the real-time wind speed value and the water mist particle size parameter in the wind speed-particle size coordinated control instruction, perform a proportional integral operation on the wind speed value and the particle size parameter, use the water vapor linkage control coefficient as a dynamic weight factor, construct a linear combination relationship between wind speed and particle size, perform a boundary check on the calculation result and the input parameter range of the evaporative cooling unit, and generate a coordinated control parameter;

[0050] The integral time constant of the proportional integral operation is set to 5 seconds, and the boundary check range is wind speed 0-15m / s and particle size 10-100μm;

[0051] S502: calling the water vapor linkage control coefficient in the coordinated control parameter, using a threshold comparator to compare the coefficient with a preset threshold, and when the coefficient exceeds the threshold, extracting the dynamic adjustment coefficient of the current wind speed value and the proportional adjustment parameter of the particle size value, performing a multiplication operation of the wind speed value and a proportional reduction operation of the particle size value, and generating a control activation signal;

[0052] The hysteresis interval of the threshold comparator is set to ±0.05 to avoid frequent switching caused by parameter fluctuations;

[0053] S503: Based on the actuator interface of the evaporative cooling unit, the target wind speed and target particle size in the control activation signal are converted into device drive instructions, the real-time value of the blade surface temperature difference sensor is synchronously collected, and the absolute difference between the temperature difference value and the baseline value is calculated. When the difference does not meet the preset convergence standard, the wind speed and particle size outputs are cyclically corrected according to the adjustment step size to generate a temperature difference baseline compliance status;

[0054] The preset convergence standard is that the absolute temperature difference is less than 0.5°C, the adjustment step is the wind speed change ±0.2m / s, and the particle size change ±1μm, and closed-loop convergence is achieved through PID control.

[0055] A smart farm environment control system based on big data analysis, wherein the smart farm environment control system based on big data analysis is used to execute the above-mentioned smart farm environment control method based on big data analysis, and the system includes:

[0056] The data acquisition and fusion module is used to obtain the upper surface temperature value and the instantaneous temperature value of the lower surface of the blade through the infrared temperature sensor, obtain the instantaneous value of the stomatal conductance through the stomatal conductance sensor, and obtain the air cavity water vapor pressure difference value through the water vapor pressure sensor. The above parameters are input into the multi-source data fusion algorithm to align the time series and remove noise, generate a dynamic physiological data set, and transmit the dynamic physiological data set to the temperature difference increase determination module and the water vapor linkage analysis module;

[0057] The temperature difference increase determination module is used to perform difference calculation on the instantaneous temperature value of the upper surface of the leaf and the instantaneous temperature value of the lower surface of the leaf using a dynamic physiological data set, compare the temperature difference values of adjacent cycles, input the temperature difference sequence into the time series trend analysis to determine the temperature difference increase. If the temperature difference increase exceeds a preset threshold for three consecutive cycles, a transpiration activity enhancement signal is generated and transmitted to the wind speed and particle size control module;

[0058] a wind speed and particle size control module, configured to increase the wind speed of the evaporative cooling unit by 0.5 m / s per cycle and decrease the water mist particle size by 5% per cycle using the transpiration activity enhancement signal, generate a wind speed-particle size coordinated control instruction, and transmit the wind speed-particle size coordinated control instruction to the dynamic feedback execution module;

[0059] a water vapor linkage analysis module, configured to perform a Pearson correlation calculation on the instantaneous value of stomatal conductance and the air cavity water vapor pressure difference value using the dynamic physiological data set; if the absolute value of the correlation coefficient is greater than 0.8, input the instantaneous value of stomatal conductance and the air cavity water vapor pressure difference value into a multivariate linear regression model to assign a weight coefficient, generate a water vapor linkage control coefficient, and transmit the water vapor linkage control coefficient to the dynamic feedback execution module;

[0060] A dynamic feedback execution module is used to integrate the wind speed-particle size coordinated control instruction and the water vapor linkage control coefficient, and input the above parameters into the feedback control algorithm to dynamically map the control parameters of the evaporative cooling unit. When the water vapor linkage control coefficient is greater than 0.6, the wind speed is increased to 1.2 times the current value and the water mist particle size is reduced to 90% of the current value. Adjustments are continuously made until the temperature difference on the blade surface returns to the set baseline.

[0061] Compared with the prior art, the advantages and positive effects of the present invention are:

[0062] In the present invention, a real-time dynamic physiological data set is constructed by collecting leaf temperature differences, stomatal conductance, and air cavity water vapor pressure differences. The temperature difference amplification signal is extracted, and the transpiration enhancement response is triggered in combination with trend analysis. Based on this signal, the wind speed is increased and the water mist particle size is decreased, realizing the coordinated regulation of particle size and wind speed. Correlation analysis introduces multivariate linear regression to construct a water vapor regulation weight coefficient. In feedback control, the wind speed and particle size are dynamically adjusted according to the weight to ensure that the leaf temperature difference returns to the baseline state. This solution realizes fine-grained regulation based on crop status, improves the accuracy and physiological adaptability of the regulation response, and enhances the system's adaptive regulation capabilities to changing environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 Schematic diagram of the workflow of the present invention.

[0064] Figure 2 Flowchart of the steps for acquiring dynamic physiological data sets of the present invention.

[0065] Figure 3 This is a flow chart of the steps for obtaining the transpiration activity enhancement signal of the present invention.

[0066] Figure 4 This is a flow chart of the steps for obtaining wind speed-particle size coordinated control instructions of the present invention.

[0067] Figure 5 This is a flow chart of the steps for obtaining the water vapor linkage control coefficient of the present invention.

[0068] Figure 6 This is a flow chart of the steps of S5 of the present invention. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0070] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0071] Example 1

[0072] See also Figure 1 The present invention provides a technical solution: a smart farm environment control method based on big data analysis, comprising the following steps:

[0073] S1: The instantaneous temperature values of the upper and lower surfaces of the leaves are obtained through infrared temperature sensors, the instantaneous stomatal conductance value is obtained through stomatal conductance sensors, and the air cavity water vapor pressure difference value is obtained through water vapor pressure sensors. A multi-source data fusion algorithm is used to align the time series and remove noise to generate a dynamic physiological data set.

[0074] S2: Perform difference calculation on the upper and lower surface temperatures of leaves in the dynamic physiological data set, compare the temperature difference values of adjacent cycles, and use time series trend analysis to determine the temperature difference increase. If the increase exceeds the threshold for three consecutive cycles, an enhanced transpiration signal is output;

[0075] S3: Based on the transpiration activity enhancement signal, the wind speed of the evaporative cooling unit is adjusted to increase by 0.5 m / s per cycle and the water mist particle size is reduced by 5% per cycle, generating a wind speed-particle size coordinated control instruction;

[0076] S4: Perform Pearson correlation calculation on stomatal conductance and air cavity water vapor pressure difference in dynamic physiological data sets. If the absolute value of the correlation coefficient is greater than 0.8, a multivariate linear regression model is used to assign weight coefficients to generate water vapor linkage control coefficients.

[0077] S5: Integrate the wind speed-particle size coordinated control instructions and the water vapor linkage control coefficient, and use the feedback control algorithm to dynamically map them to the evaporative cooling unit. When the water vapor linkage control coefficient is greater than 0.6, the wind speed is increased to 1.2 times the current value, and the water mist particle size is reduced to 90%. Continue to adjust until the temperature difference on the blade surface returns to the set baseline.

[0078] The dynamic physiological data set specifically includes the instantaneous temperature value of the upper surface of the leaf, the instantaneous temperature value of the lower surface of the leaf, the instantaneous value of the stomatal conductance, and the water vapor pressure difference value of the air cavity. The transpiration activity enhancement signal includes the temperature difference increase, the number of consecutive cycles, and the temperature difference threshold. The wind speed-particle size coordinated control instructions include the wind speed adjustment gradient and the particle size adjustment gradient. The water vapor linkage control coefficient specifically refers to the correlation coefficient threshold and the regression weight. The control parameters dynamically mapped by the feedback control algorithm include the wind speed multiple and the particle size percentage.

[0079] Multi-source data fusion algorithms include timestamp alignment algorithm, sliding window standard deviation calculation method, wavelet threshold denoising method and weighted Kalman filter algorithm;

[0080] The wind speed adjustment gradient and the particle size adjustment gradient are parameter solutions after optimizing the transpiration suppression efficiency based on the multivariate regression model;

[0081] The threshold for the absolute value of the correlation coefficient is the critical value verified by the t-test at a significance level of α = 0.01 and degrees of freedom of n-2;

[0082] The feedback control algorithm is a dynamic parameter mapping rule based on proportional-integral-differential, with the proportional coefficient set to 0.8, the integral time constant to 10 seconds, and the differential time constant to 2 seconds.

[0083] See also Figure 2 ,The steps for acquiring dynamic physiological data sets are as follows:

[0084] S101: Collect the instantaneous values of the upper and lower surface temperatures of the blade output by the infrared temperature sensor, synchronously obtain the instantaneous conductance value output by the stomatal conductance sensor and the instantaneous value of the water vapor pressure difference output by the water vapor pressure sensor, and use a timestamp alignment algorithm to match the four types of parameters according to millisecond-level timestamps to generate synchronized time series parameters;

[0085] The timestamp alignment algorithm uses the Coordinated Universal Time (UTC) time base to convert the timestamps of the sensor-collected parameters into millisecond-level UTC time.

[0086] This step aims to obtain the original readings of multiple physiological parameters of plant leaves at specific time points and accurately align them in time. First, data is collected by sensors deployed at key positions on the leaves. The specific operation is as follows: non-contact infrared temperature sensor A is used to monitor the instantaneous temperature value of the upper surface of the leaf in real time, while infrared temperature sensor B is used to monitor the instantaneous temperature value of the lower surface of the leaf. The two sensors work synchronously to obtain temperature readings. A specific sampling time is 10:00:00 100 milliseconds on April 24, 2025. The reading of sensor A is T upper = 28.5℃, there is a slight delay of 102 milliseconds, and the sensor B reading is T lower =28.2℃; at the same time, the probe of the contact stomatal conductance sensor is fixed on the leaf surface to measure the instantaneous value of stomatal conductance G s , the reading is 0.25 mol·m at 98 milliseconds -2 ·s -1 ; And use the water vapor pressure sensor to measure the difference between the saturated water vapor pressure inside the blade and the water vapor pressure of the ambient air around the blade, that is, the instantaneous value of the water vapor pressure difference V p , a reading of 1.8 kPa is obtained at 105 milliseconds.

[0087] After obtaining these discrete data with their own timestamps, the timestamp alignment algorithm is used to ensure that all parameters reflect the physiological state at the same moment. Set an alignment reference time point and select the timestamp that arrives earliest or is closest to a regular time point among all sampling timestamps. Here, 100 milliseconds is used as the reference time. Set a time tolerance window. The size of this window is determined by the sensor sampling frequency and data transmission delay. It is set to ±5 milliseconds, that is, the time interval of [95ms, 105ms]. Check the latest sampling data of each sensor within this time window: G s At 98ms (falling into the window), T upper In 100ms (falling into the window), T lower At 102ms (falling into the window), V p At 105ms (falling into the window). All key parameters are sampled within this window. Applying the nearest neighbor alignment method, all parameter values in the window closest to the reference timestamp 100ms are uniformly attributed to the time point of 100ms. Thus, the first synchronization timing parameter point is generated: (timestamp = 100ms, T upper =28.5℃,T lower =28.2℃,G s =0.25mol·m -2 ·s-1 ,V p =1.8kPa). This acquisition and alignment process is repeated continuously, and more synchronized data points are subsequently acquired at time points such as 200ms and 300ms, such as (timestamp = 200ms, T upper =28.6℃,T lower =28.3℃,G s =0.26mol·m -2 ·s-1,V p =1.85kPa), ultimately forming a time series containing synchronized physiological parameters at multiple time points.

[0088] S102: Based on the synchronization time series parameters, the sliding window standard deviation calculation method is used to remove outliers from the temperature value sequence, and the wavelet threshold denoising is performed on the stomatal conductance value sequence and the water vapor pressure difference value sequence. The high-frequency noise components are truncated and the low-frequency features are retained to generate a reconstructed feature sequence;

[0089] Wavelet threshold denoising uses the Stein unbiased risk estimation method to determine the threshold and cut off the noise component with a frequency higher than 1kHz;

[0090] After obtaining the synchronization timing parameter sequence, it needs to be preprocessed to eliminate noise and abnormal values. upper and T lower ), the sliding window standard deviation calculation method is used to remove outliers. The sliding window size is set to 11 data points. The selection of this window size is intended to balance the noise filtering and smoothing effects with the preservation of signal details. For the i-th temperature value T in the sequence upper,i , select the data segment containing this point and the 5 data points before and after it (a total of 11 points) to form the analysis window. Calculate the arithmetic mean μ and standard deviation σ of these 11 temperature values. Set the judgment standard of the outlier to exceed the mean by 3 times the standard deviation, that is, |T upper,i -μ|>3σ. The selection of 3σ as the threshold is based on the assumption of normal distribution. The probability of data points outside this interval appearing is extremely low and is usually regarded as anomalies. If a temperature value, such as 31.2℃, is within a window, the window average is μ=28.81℃, and the standard deviation is σ=0.80℃, then the judgment threshold is 3σ=2.40℃. Calculate the deviation of this point |31.2-28.81|=2.39℃. Since 2.39≤2.40, this point is not judged as an outlier. If another value in the window is 32.0℃, its deviation is |32.0-28.81|=3.19℃. Since 3.19>2.40, this point is judged as an outlier and its value is replaced with the window average of 28.81℃. For T upper and T lower This sliding window detection and replacement operation is performed on each data point in the sequence.

[0091] For the stomatal conductance value sequence G s and the water vapor pressure difference sequence V p , using the wavelet threshold denoising method. The Daubechies4 ('db4') wavelet basis is selected because it has good time-frequency localization and regularity, which is suitable for processing such physiological signals. The number of decomposition layers is set to 5, which is selected based on the main frequency components and noise characteristics of the signal and is sufficient to separate most high-frequency noise. s The sequence performs wavelet decomposition to obtain the wavelet coefficients of each layer (approximate coefficients and detail coefficients). The standard deviation of the noise is estimated by the median absolute deviation (MAD) of the first layer detail coefficient cD1 noise =MAD(cD1) / 0.6745. This method is not sensitive to outliers in the data and is relatively robust. Get the estimated value of the noise standard deviation, such as σ noise =0.01mol·m-2·s -1 Calculate the threshold based on the Universal Threshold rule (UniversalThreshold) Where N is the length of the signal sequence. Assume N = 10,000 data points, and calculate Use the soft threshold function to process the detail coefficients d of the 1st to 5th layers: if |d|≤thr, set the coefficient to zero new =0; if |d|>thr, the coefficient shrinks to d new =sign(d)(|d|-thr). This processing method can remove noise while maintaining the smoothness of the signal. The processed detail coefficients and the approximate coefficients of the bottom layer (layer 5) are retained. These coefficients represent the main low-frequency characteristics of the signal. Finally, the inverse wavelet transform is performed using these processed coefficients to reconstruct the denoised G s Sequence. p The same denoising process is performed on the sequences. Through the above processing, a reconstructed feature sequence with significantly reduced noise level is generated, which better reflects the true physiological change trend.

[0092] S103: Calling the reconstructed feature sequence, calculating the instantaneous value of the temperature difference between the upper and lower surfaces of the leaf, combining the instantaneous value of stomatal conductance and the instantaneous value of water vapor pressure difference, and dynamically fusing the multi-dimensional parameters using the weighted Kalman filter algorithm. The fusion weights are assigned according to the covariance matrix to generate a dynamic physiological data set;

[0093] The fusion weights of the weighted Kalman filter algorithm are assigned according to the normalized values of the diagonal elements of the covariance matrix.

[0094] Call the reconstructed feature sequence (ie, the denoised T upper (t),Tlower (t),G s (t),V p (t)), multi-source information fusion is performed to obtain a more accurate and reliable dynamic physiological state assessment. First, the instantaneous temperature difference between the upper and lower surfaces of the leaf is calculated. For each time point t, T diff (t) = T upper (t)-T lower (t). Using the data at time point t = 100ms, after denoising, T upper (100ms)=28.5℃,T lower (100ms) = 28.2°C, and T diff (100ms)=28.5-28.2=0.3℃.

[0095] Next, the calculated temperature difference T diff (t) and the denoised stomatal conductance G s (t) and the vapor pressure difference V p (t) is used as input and the weighted Kalman filter algorithm is used for dynamic fusion. This algorithm is suitable for fusing multi-sensor data with noise and can estimate the state of the system online. The state vector of the system is defined as x k =[T diff,k ,G s,k ,V p,k ] T , where k represents the discrete time step. The measurement vector is That is, the direct observation value at time k after S102 processing. Set the state transfer matrix F to the unit matrix I, which means that the physiological state is continuous in a short time (one sampling interval) without drastic mutations. The observation matrix H is also set to the unit matrix I, which means that each variable in the state vector can be directly measured. Define the process noise covariance matrix Q, whose diagonal elements represent the uncertainty of the model prediction or the degree of natural fluctuation of the state over time. Set Q = diag (0.005 2 ,0.01 2 ,0.05 2 ), indicating that the natural fluctuation of temperature difference is the smallest and the fluctuation of water vapor pressure difference is the largest. Define the measurement noise covariance matrix R, whose diagonal elements are the estimated values of the measurement error variance of each sensor, reflecting the reliability of the measurement, and set R = diag(0.02 2 ,0.03 2 ,0.1 2 ), indicating that the temperature difference measurement has the highest accuracy (small variance) and the water vapor pressure difference measurement has the lowest accuracy (large variance).

[0096] The core of weighted Kalman filtering is to calculate the Kalman gain in is the prediction error covariance. This gain matrix determines the effect of the new measurement value z on the updated state estimate. k and the predicted value at the previous moment The smaller diagonal elements in R (corresponding to high-precision measurements) will result in the corresponding state variables being k The algorithm iterates the prediction step. and update step Through continuous iteration, the Kalman filter can effectively fuse multi-source information, suppress noise, and output a set of time series x that is smoother and closer to the real physiological state. k , this set of sequences is the final generated dynamic physiological data set.

[0097] Table 1. Example of dynamic physiological dataset segments

[0098]

[0099] As shown in Table 1, the table lists a segment of the dynamic physiological data set generated after weighted Kalman filtering, which contains the estimated values of the temperature difference between the upper and lower surfaces of the leaf, stomatal conductance, and air cavity water vapor pressure difference corresponding to the fused time stamps from 100ms to 1000ms.

[0100] See also Figure 3 The specific steps for obtaining the transpiration activity enhancement signal are as follows:

[0101] S201: Obtain the upper and lower surface temperature data of each leaf in the dynamic physiological data set, use the sampling time as the index, align the upper and lower surface temperatures of the same leaf at the same time along the time axis, perform subtraction operations on each time point, calculate the temperature difference between the two surfaces of the leaf per unit time, and generate a temperature difference sequence corresponding to multiple time points;

[0102] The unit time is the sensor sampling interval, which is set to 100 milliseconds;

[0103] The purpose of this step is to calculate the indicator reflecting the intensity of leaf transpiration activity: the temperature difference between the upper and lower surfaces of the leaf. The denoised upper surface temperature T is obtained from the reconstructed feature sequence processed by S102. upper (t) and the lower surface temperature T lower (t) time series data. Using sampling time t as index, ensure that at each time point, the T corresponding to the same leaf is upper (t) and T lower (t) Perform pairing.

[0104] For each time point t, perform subtraction operation to calculate the instantaneous temperature difference value T diff (t) = Tupper (t)-T lower (t). For the time point t = 300ms, if the upper surface temperature after denoising is 28.7°C and the lower surface temperature is 28.38°C, then the temperature difference at this moment is calculated as T diff (300ms) = 28.7 - 28.38 = 0.32°C. Repeat this calculation process for all time points in the entire time series (e.g., 100ms, 200ms, …, 1000ms, …). If you use the temperature difference data fused by S103 directly (as shown in Table 1), you can simply extract this column of data. The result is a chronological sequence of temperature differences, such as [0.31, 0.30, 0.32, 0.33, 0.34, 0.35, 0.36, 0.35, 0.37, 0.38, …], in °C. This sequence reflects the differences in the cooling effect caused by transpiration on the leaves at different times.

[0105] S202: Based on the temperature difference sequence, data segments of two adjacent cycles are intercepted according to the period division rule. For the temperature difference at each time point, the operation of subtracting the value of the previous cycle from the value of the next cycle is performed. The change in the temperature difference at the corresponding time point between the cycles is calculated. All the changes are arranged in chronological order to generate a sequence of temperature difference increases between adjacent cycles.

[0106] The cycle division rule is to define every 300 consecutive sampling points as a control cycle;

[0107] Based on the temperature difference sequence T diff (t), analyzing its periodic changes to determine the trend of transpiration activity. The analysis period was set to 5 minutes, based on the typical time scale of plant physiological activity in response to environmental changes. If the sampling frequency is 10 Hz (one point every 100 ms), each period contains 5 × 60 × 10 = 3000 data points.

[0108] According to the length of this cycle, the continuous temperature difference sequence is divided into multiple data segments. Select two adjacent cycles and record them as the kth cycle and the k+1th cycle. Let the data sequence of the kth cycle be T diff,k (i) The data sequence of the k+1th cycle is T diff,k+1 (i), where i is the index of the time point within the cycle, ranging from 1 to 3000. Subtract the data points at the corresponding positions (same index i) in the two sequences and calculate the change in the temperature difference at the same time between the cycles: ΔT diff (i) = T diff,k+1 (i)-T diff,k (i) Take a specific calculation example: for the 100th time point in the cycle (corresponding to 10 seconds after the start of the cycle), if the temperature difference T at this point in the kth cycle is diff,k(100) = 0.35 °C, and the temperature difference T corresponding to the k+1 period diff,k+1 (100) = 0.38°C, then the temperature difference change at this time point is calculated as ΔT diff (100) = 0.38 - 0.35 = 0.03 °C. This subtraction operation is performed for all 3000 time points (i from 1 to 3000) in the cycle. All 3000 calculated changes ΔT diff (i) Arrange the data in chronological order (i.e., index i from 1 to 3000) to form a new sequence called the adjacent cycle temperature difference increase sequence. This sequence reflects the change in temperature difference at each moment relative to the previous cycle within a complete cycle.

[0109] S203: Call the temperature difference amplification sequence of adjacent periods, take three consecutive periods as the window unit, perform a summation operation on the amplification values of multiple time points in the window, compare the summation result with the preset transpiration activity amplification threshold item by item, and generate a transpiration activity enhancement signal when the cumulative amplification of any window exceeds the threshold.

[0110] The threshold for transpiration activity increase is the cumulative increase range of the temperature difference calibrated by the gradient descent method optimization experiment, ranging from 3.2°C to 4.5°C;

[0111] Call the adjacent cycle temperature difference amplification sequence ΔT diff (i) (i = 1 to 3000) is used to determine whether transpiration is significantly enhanced. A sliding window method is used, and the window covers three consecutive cycles to compare the results, that is, it is necessary to analyze the increase sequence ΔT of period k and k+1 diff,k→k+1 (i) Amplification sequence ΔT of periods k+1 and k+2 diff,k+1→k+2 (i) and the amplification sequence ΔT of periods k+2 and k+3 diff,k+2→k+3 (i).

[0112] Perform a sum operation on all the increment values in each increment sequence. Calculate the total increment from the kth period to the k+1th period. Similarly, calculate Sum k+1→k+2 and Sum k+2→k+3. This summation result represents the cumulative change in temperature difference during the entire cycle relative to the previous cycle. A transpiration activity increase threshold is set to determine whether the cumulative increase has reached a significant level. The basis for setting this threshold is: through previous experimental observations and data analysis, when plants enter a vigorous transpiration state, the cumulative increase in the temperature difference of their leaves within a 5-minute cycle usually exceeds a specific value. This threshold is set to 60℃·point (the unit is degrees Celsius multiplied by the number of points, representing the total change of 3000 points). This value of 60 is obtained by analyzing a large amount of leaf data under normal growth and high transpiration conditions, and selecting a cumulative increase value that can effectively distinguish between the two states.

[0113] The calculated total increase Sum between each cycle k→k+1 ,Sum k+1→k+2 ,Sum k+2→k+3 Compare each with the threshold value 60. Check Sum k→k+1 >60 whether it is true, Sum k+1→k+2 >60 is true, and Sum k+2→k+3 >60. As long as any one (or more) of the total increases in the comparison between these three consecutive cycles exceeds the threshold of 60, it is determined that the transpiration activity has increased significantly. For example, if the Sum k+1→k+2 =75. Since 75 > 60, the condition is met. At this point, the system generates a transpiration activity enhancement signal, which can be a Boolean value (true) or a flag (set to 1). If the total increase in all three cycles does not exceed 60, the signal is not generated (the signal value is false or the flag is 0).

[0114] See also Figure 4 The specific steps for obtaining wind speed-particle size coordinated control instructions are as follows:

[0115] S301: Based on the transpiration activity enhancement signal, the current wind speed value is called, and the initial wind speed and the fixed step length parameter are input into the periodic increasing function by adding the incremental step length to the current value. The superposition value of the wind speed in each cycle is calculated to generate the wind speed increment value;

[0116] The incremental step parameter is calibrated by the fluid mechanics model to calibrate the wind duct resistance coefficient of the evaporative cooling unit. The step unit is m / s. 2 ;

[0117] When the transpiration active enhancement signal is generated (flag bit is 1), the control instruction generation process is started. First, the current real-time wind speed value V of the system is obtained. current This value is provided by the current operating status of the wind turbine recorded by the control system and is 0.6m / s. The wind speed is gradually increased in subsequent control cycles by using the incremental step superposition method.

[0118] Set a fixed wind speed increase step size ΔV = 0.15m / s. The selection of this step size is based on the control objectives and equipment capabilities, aiming to achieve smooth and effective wind speed adjustment, avoiding excessive changes that cause stress to plants or fail to achieve the expected control effect. Set the control cycle to be consistent with the analysis cycle in S202, which is 5 minutes. Use a cycle increment function to calculate the target wind speed values for the next few cycles. The specific calculation method is: the target wind speed V1 of the first control cycle (i.e., the next 5-minute cycle after receiving the enhanced signal) =

[0119] V current +ΔV=0.6+0.15=0.75m / s. The target wind speed of the second control cycle V2=V1+ΔV=0.75+0.15=0.90m / s. The target wind speed of the third control cycle V3=V2+ΔV=0.90+0.15=1.05m / s. This calculation continues until the preset wind speed limit is reached or a command to stop control is received. This generates a sequence of target values with increasing wind speed.

[0120] [0.75, 0.90, 1.05, 1.20, 1.35, …], all in m / s.

[0121] S302: Divide the nodes according to the period of the wind speed increment value, extract the median value of the original water mist particle size distribution, reduce the upper and lower limits of the particle size range simultaneously in a geometrically decreasing manner, and compress and translate the particle size distribution curve using a segmented iterative algorithm to generate a particle size decreasing gradient;

[0122] The compression and translation ratio of the segmented iterative algorithm is to reduce the width of the particle size distribution interval by 5% per cycle;

[0123] The particle size distribution of the mist is adjusted synchronously with the increasing wind speed. At the beginning of each control cycle (synchronized with the wind speed adjustment cycle, i.e., one node every 5 minutes), the current or original particle size distribution (PSD) of the mist is first obtained. The median value D50 of the distribution (i.e., the diameter below which 50% of the total volume or number of particles are present) is used as the primary characterization parameter. The current PSD characteristic recorded by the system is D50 = 60μm.

[0124] The characteristic parameters of the particle size distribution are reduced synchronously in a geometrically decreasing manner to produce finer water mist and enhance the evaporative cooling effect. A decreasing proportional factor f = 0.92 is set, which means that the characteristic value of the particle size in each cycle is reduced to 92% of the original. The selection of this factor takes into account the capacity of the atomization equipment and the expected cooling efficiency improvement rate. At the same time, not only D50 is adjusted, but also the range of the particle size distribution. Take the interval [40μm, 80μm] defined by the original distribution D10 (10% of the particle diameter is less than this value) and D90 (90% of the particle diameter is less than this value) as an example. In the first control cycle (corresponding to the target wind speed V1 = 0.75m / s), calculate the new particle size parameter: the new median value D 50,1 =D 50,0 ×f=60×0.92=55.2μm. New lower limit of distribution range D lower,1 =D lower,0 ×f=40×0.92=36.8μm. New upper limit of distribution range D upper,1 =D upper,0 ×f=80×0.92=73.6μm.

[0125] The actual control parameters (such as nozzle pressure, frequency, etc.) are adjusted using a segmented iterative algorithm to make the generated particle size distribution as close as possible to the new target distribution characteristics (D50 = 55.2 μm, interval

[0126] =[36.8μm,73.6μm]). For each subsequent control cycle, the target particle size parameter of the previous cycle is multiplied by the factor f to calculate the new target parameter. The target D of the second cycle is 50,2 =D 50,1 ×f=55.2×0.92=50.784μm, and the interval is simultaneously reduced to [33.9μm, 67.7μm]. Continuing this process, a series of target particle size distribution parameters that decrease over time are generated, forming a particle size decreasing gradient:

[0127] [(D50=55.2,Range=[36.8,73.6]),(D50=50.8,Range=[33.9,67.7]),(D50=46.7,Range=[31.1,62.3]),…], all units are μm.

[0128] S303: On the control cycle time axis, a Cartesian product mapping is performed on the numerical sequence of the wind speed increment value and the distribution parameter of the particle size decrement gradient to establish a two-dimensional parameter combination matrix of wind speed and particle size. An instruction parameter set is generated through matrix coordinate transformation to obtain a wind speed and particle size coordinated control instruction;

[0129] The matrix coordinate transformation rule is that the wind speed-particle size coupling coefficient is 0.7, and the linear correlation between wind speed and particle size is calibrated through experiments.

[0130] On the time axis of the control cycle (with 5 minutes as a step), the wind speed increment target value sequence

[0131] [0.75, 0.90, 1.05, …] m / s and the particle size decreasing gradient parameter sequence generated by S302

[0132] [(D50=55.2,Range=[36.8,73.6]),(D50=50.8,Range=[33.9,67.7]),(D50=46.7,Range=[31.1,62.3]),…]μm for synchronous pairing.

[0133] Specifically, for the first control cycle, the target parameter combination is (target wind speed = 0.75m / s, target particle size characteristics = {D50 = 55.2μm, Range = [36.8, 73.6]μm}). For the second control cycle, the target parameter combination is (target wind speed = 0.90m / s, target particle size characteristics

[0134] ={D50=50.8μm,Range=[33.9,67.7]μm}). Similarly, the corresponding relationship between time, target wind speed, and target particle size parameters is established.

[0135] Convert these paired target parameter groups into an instruction format that the control system can recognize and execute. This involves converting descriptive parameters of the particle size distribution (such as D50 and range) into specific equipment control set values (such as nozzle pressure, flow, etc.), and converting the target wind speed into a control signal for the fan (such as voltage, frequency or PWM duty cycle). Generate an instruction parameter set in the form of a list or table, where each row contains the index of the control cycle, the target wind speed value for that cycle, and the corresponding target particle size control parameter.

[0136] Table 2 Example of wind speed-particle size coordinated control instruction parameter table

[0137]

[0138] Table 2 shows the first five cycles of the generated wind speed-particle size coordinated control instruction parameter set, clearly outlining the target wind speed and water mist particle size distribution characteristics to be achieved in each control cycle. This instruction set will guide the specific operation of the evaporative cooling unit.

[0139] See also Figure 5 , the specific steps for obtaining the water vapor linkage control coefficient are:

[0140] S401: detecting stomatal conductance and air cavity water vapor pressure difference in the dynamic physiological data set, and calculating the covariance and standard deviation ratio of the stomatal conductance time series change value and the air cavity water vapor pressure difference time series change value based on the stomatal conductance time series change value and the air cavity water vapor pressure difference time series change value to generate a Pearson correlation coefficient value;

[0141] In order to quantify the linkage between stomatal conductance and water vapor pressure difference, it is necessary to calculate the Pearson correlation coefficient between them. From the dynamic physiological data set (refer to Table 1 and subsequent data), the stomatal conductance G of the most recent period (set to 15 minutes, if the sampling frequency is 10Hz, it contains 15×60×10=9000 data points) is extracted. s and the air cavity water vapor pressure difference V p time series data.

[0142] Based on these 9000 pairs of data points (G s,i ,V p,i ), calculate the respective mean and standard deviation. The calculation process is: G s Average value V p Average value Where N = 9000. Calculate the sample standard deviation: Calculated Next, calculate the covariance between the two: Calculate Cov(G s ,V p )=-0.0045.

[0143] Finally, calculate the Pearson correlation coefficient r, which is calculated as follows: Substituting the calculated values: The calculated result, r = -0.75, is the Pearson correlation coefficient between stomatal conductance and vapor pressure difference at the current stage, indicating a strong negative correlation between the two.

[0144] S402: calling the Pearson correlation coefficient value, comparing the absolute value of the correlation coefficient with a preset correlation threshold, and if the condition is met, establishing a linear combination equation for the stomatal conductance and the air cavity water vapor pressure difference using the least squares method to generate a weight distribution benchmark parameter;

[0145] Using the Pearson correlation coefficient r=-0.75, we can judge G s and V p Whether the linear correlation strength between them meets the preset standard. Set the correlation threshold r threshold=0.5. This threshold is set based on statistical significance considerations: for a sample size of N = 9000, a correlation coefficient greater than 0.5 usually means that the data are highly statistically significant (the p-value is much less than 0.01), indicating that the linear relationship between the two is very clear and it is worth establishing a linear model.

[0146] The absolute value of the calculated Pearson correlation coefficient |r|=|-0.75|=0.75 is compared with the threshold. Since 0.75>0.5, the condition is met. Therefore, it can be considered that G s With V p If there is a significant linear relationship between the two, proceed to the next step. If |r| is not greater than 0.5, the linear relationship is considered insignificant and a different model or adjustment strategy may be needed.

[0147] When the conditions are met, the least square method is used to establish G s V p The linear regression equation G s =

[0148] aV p + b. The statistics calculated in S401 are used to determine the regression coefficients a (slope) and b (intercept).

[0149] Slope intercept The resulting linear equation is

[0150] G s =-0.2V p This equation, along with the correlation coefficient r = -0.75, constitutes the baseline parameters for weight assignment, which quantify the sensitivity of stomatal conductance to changes in vapor pressure deficit.

[0151] S403: Call the weight distribution benchmark parameters, extract the difference between the stomatal conductance and the air cavity water vapor pressure at adjacent time points, integrate the ambient temperature sensor acquisition value and the leaf surface area measurement value, and use the formula:

[0152]

[0153] Calculate the ratio of the square root of the dynamic change difference to the weighted sum of the ambient temperature, and combine it with the regression model residual correction to generate the water vapor linkage control coefficient;

[0154] Where W represents the water vapor linkage control coefficient, ΔG s is the difference in stomatal conductance at adjacent time points, ΔV p is the difference in water vapor pressure difference between adjacent time points in the air cavity, λ i is the ambient temperature sensor calibration coefficient, is the real-time measurement value of the temperature sensor, n is the number of sensors, r is the Pearson correlation coefficient, S leaf is the calibrated value of the blade surface area;

[0155] Ambient temperature sensor calibration coefficient λ i The leaf surface area calibration value S is obtained by least square fitting the nonlinear relationship between ambient temperature and stomatal conductance. leaf is the mean value of the measured leaf projected area.

[0156] The weight distribution benchmark parameters (mainly the correlation coefficient r = -0.75) are called, and the latest adjacent time point data are extracted from the dynamic physiological data set to calculate the instantaneous change. The data points at t = 500ms and t = 400ms in Table 1 are selected: G s (500ms)=0.268,V p (500ms)=1.88;G s (400ms)=0.265,V p (400ms) = 1.87. Calculate the difference between adjacent time points: ΔG s =G s (500ms)-G s (400ms)=0.268-0.265=0.003mol·m -2 ·s -1 , ΔV p =V p (500ms)-V p (400ms)=1.88-1.87=0.01kPa.

[0157] At the same time, the real-time collection value of the ambient temperature sensor is integrated. The system is equipped with n=3 ambient temperature sensors, which are placed at different locations in the target area. The real-time readings obtained are Each sensor is calibrated to obtain its own calibration coefficient λ i , used to correct the measurement deviation, set λ1 = 1.01, λ2 = 0.99, λ3 = 1.00. These coefficients are unitless proportional factors. In addition, obtain the calibration value S of the blade surface area leaf This value is the average surface area of a representative leaf measured over a 65cm 2 .

[0158] The water vapor linkage control coefficient W is calculated using the following formula: In this formula: ΔG s and ΔV p represent the instantaneous changing rates of stomatal conductance and water vapor pressure difference, respectively. A measure representing the average ambient temperature or total heat level after calibration. |r| is the strength of the correlation between stomatal conductance and vapor pressure deficit. S leaf is the characteristic size parameter of the blade. leaf ) is the natural logarithm of the leaf area.

[0159] Before substituting into calculations, it is necessary to unify the units of physical quantities. s The unit is mol·m -2 ·s -1 , ΔG s The units are the same. ΔV p The unit is kPa. The unit is °C. i and r are unitless. ΔG s ΔV p The unit is mol·m -2 ·s -1 kPa. Ambient temperature summation term The unit is °C. Leaf area S leaf The unit of measurement is cm 2 , but other parameters in the formula involve meters (m), so for consistency, the leaf area needs to be converted to square meters (m 2 ). Set conversion rules: 1m 2 =10,000cm 2 Therefore, it is necessary to convert the 2 Divide the unit value by 10000. ln(S leaf ) is unitless. Note that in the numerator (unit )and (unit ℃) are directly added. In the physical sense, the direct addition of quantities of different units needs to be handled with caution. The intention of the formula design here is to combine the dynamic intensity of physiological changes (the former) and the environmental heat load (the latter) as the total signal intensity of driving regulation. In actual systems, these two quantities will be normalized or converted by coefficients that are not explicitly written before addition, so that their numerical sizes and relative importance performance are appropriately combined. Here, according to the formula structure, the final W is regarded as a comprehensive, dimensionless or composite unit regulation index.

[0160] Substitute the values for calculation: The first term in the numerator: |r|=|-0.75|=0.75 The logarithm calculation in the second term of the denominator: ln(S leaf )=ln(0.0065)≈-5.036. The second term in the denominator: 1+ln(S leaf)=1+(-5.036)=-4.036 Denominator=|r|·(1+ln(S leaf ))=0.75×(-4.036)=-3.027

[0161] Final calculation The calculated value W≈-26.89 is negative and has a large absolute value. This result indicates that under the current conditions (high ambient temperature, some fluctuations in physiological parameters, and a strong negative correlation between Gs and Vp), stronger regulatory measures tending to lower leaf temperature or inhibit transpiration are needed.

[0162] The benefit of the formula is that it takes into account the instantaneous change rate of the leaf physiological state (ΔG s ,ΔV p ), ambient temperature (T env ), the intrinsic coupling strength between key physiological parameters (r), and the physical properties of the leaf itself (S leaf ), a dynamic, multi-factor driven control coefficient W is generated, which is more adaptable to complex and changing actual situations than the control based on a single threshold or static model.

[0163] See also Figure 6 , the specific steps of S5 are:

[0164] S501: Integrate the real-time wind speed value and the water mist particle size parameter in the wind speed-particle size coordinated control instruction, perform proportional integral operation on the wind speed value and the particle size parameter, use the water vapor linkage control coefficient as a dynamic weight factor, construct a linear combination relationship between wind speed and particle size, perform boundary verification on the calculation result and the input parameter range of the evaporative cooling unit, and generate coordinated control parameters;

[0165] The integral time constant of the proportional integral operation was set to 5 seconds, and the boundary check range was 0-15 m / s for wind speed and 10-100 μm for particle size;

[0166] The wind speed-particle size coordinated control instruction (for the target value of the current cycle, take the first cycle as an example: target wind speed V target,S303 =0.75m / s, target D50D target,S303 =55.2μm) is integrated with the water vapor linkage control coefficient W≈-26.89 calculated in S403. W is used as a dynamic weight factor to fine-tune the basic control instructions.

[0167] Set adjustment rules: The final output wind speed and particle size are adjusted based on the size and sign of W based on the basic target value. Define the adjustment coefficient: wind speed adjustment coefficient k V =0.005(m / s) / W unit, particle size adjustment coefficient k D=-0.1(μm) / W unit. The setting of these two coefficients is based on simulation or experimental optimization to determine the sensitivity and direction of W to wind speed and particle size adjustment. V A positive value indicates that the wind speed decreases when W is negative. D Negative W indicates that the particle size increases when the W value is negative (this is contrary to expectations, and the logic is corrected: negative W indicates that cooling needs to be strengthened, and the wind speed should be increased and the particle size should be reduced. Reset k V =-0.005,k D =0.1). Calculate the adjusted wind speed: V adj =V target,S303 +k V ×W=0.75+(-0.005)×(-26.89)=0.75+0.13445=0.88445m / s. Calculate the adjusted particle size: D adj =D target,S303 +k D ×W=55.2+0.1×(-26.89)=55.2-2.689=52.511μm.

[0168] The calculated adjusted parameter V adj ≈0.88m / s and D adj ≈52.5μm and the physical or set operating limits of the evaporative cooling unit are checked. The feasible range of the wind speed is set to [0.1, 2.0] m / s, and the feasible range of the water mist particle size (D50) is [20, 80] μm. Check the calculation results: 0.1≤0.88≤2.0, 20≤52.5≤80. Both parameters are within the valid range. After the verification is passed, the two values V adj and D adj As preliminary collaborative control parameters.

[0169] S502: The water vapor linkage control coefficient in the coordinated control parameters is called, and a threshold comparator is used to compare the coefficient with a preset threshold. When the coefficient exceeds the threshold, the dynamic adjustment coefficient of the current wind speed value and the proportional adjustment parameter of the particle size value are extracted, and a multiplication operation of the wind speed value and a proportional reduction operation of the particle size value are performed to generate a control activation signal.

[0170] The hysteresis interval of the threshold comparator is set to ±0.05 to avoid frequent switching caused by parameter fluctuations;

[0171] Call the water vapor linkage control coefficient W≈-26.89. Set a control activation threshold W activate_threshold =25. This threshold is set based on operational experience. When |W| exceeds this threshold, it indicates that the current physiological or environmental conditions deviate significantly from the steady state and require the initiation of a more forceful or refined adjustment strategy.

[0172] Perform a threshold comparison: Calculate |W| = |-26.89| = 26.89. Because 26.89 > 25, the absolute value of the coefficient exceeds the activation threshold. The activation condition is met.

[0173] When the coefficient exceeds the threshold, the preliminary cooperative control parameter V generated in S501 is adj and D adj Perform secondary adjustment. Set the wind speed dynamic adjustment coefficient f for exceeding the threshold value V_adjust =1.1(indicates that V adj Add 10% on the basis), and the particle size ratio adjustment parameter f D_adjust =0.95(indicates that D adj These two coefficients are preset to further enhance the control effect when strong intervention is needed (increase wind speed and reduce particle size to enhance cooling). Perform multiplication operation to make the final adjustment: Final target wind speed V final =V adj ×f V_adjust =0.88445×1.1=0.972895m / s. Final target particle size D final =D adj ×f D_adjust =52.511×0.95=49.88545μm.

[0174] Generate a control activation signal, which not only indicates that control needs to be executed, but also contains the two final target wind speed values V after secondary adjustment. final ≈0.97m / s and target particle size value D final ≈49.9μm.

[0175] S503: Based on the actuator interface of the evaporative cooling unit, the target wind speed and target particle size in the control activation signal are converted into device drive instructions, the real-time value of the blade surface temperature difference sensor is synchronously collected, and the absolute difference between the temperature difference value and the baseline value is calculated. If the difference does not meet the preset convergence standard, the wind speed and particle size outputs are cyclically corrected according to the adjustment step size to generate a temperature difference baseline compliance status;

[0176] The preset convergence standard is that the absolute temperature difference is less than 0.5°C, the adjustment step is the wind speed change ±0.2m / s, and the particle size change ±1μm, and closed-loop convergence is achieved through PID control.

[0177] The final target wind speed V is set by using the actuator interface of the evaporative cooling unit. final ≈0.97m / s and target particle size D final≈49.9μm is converted into specific device driver instructions. This means converting 0.97m / s into the voltage, frequency, or PWM signal value required to drive the fan to achieve that wind speed, and converting 49.9μm (D50) into the parameter settings for controlling the atomization system (such as adjusting water pressure and nozzle vibration frequency) to produce that particle size distribution. These instructions are then sent to the corresponding actuators (fan controller, atomization controller).

[0178] While executing the control instructions, the system continuously collects the real-time temperature difference T on the blade surface through sensors. diff,actual After a period of control operation, the collected temperature difference is T diff,actual =0.34℃. Set a target reference temperature difference T diff,baseline =0.30℃, which represents the leaf temperature difference when the plant's physiological state is ideal under this environment. Calculate the absolute difference between the current actual temperature difference and the reference temperature difference |Error| = |T diff,actual -T diff,baseline |=|0.34-0.30|=0.04℃.

[0179] Compare this difference with the preset convergence standard ∈. The convergence standard is set to ∈=0.05℃, which means that when the difference between the actual temperature difference and the target reference temperature difference is less than or equal to 0.05℃, it is considered that the control has achieved a satisfactory effect. Make a judgment: |Error|=0.04, because 0.04≤0.05, the current difference meets the convergence standard. The system determines that it is "temperature difference baseline standard status", and can maintain the current control parameters or enter the next control cycle as planned. If the difference does not meet the convergence standard, for example, if the calculation shows |Error|=0.08℃, then 0.08>0.05, and the status is "not up to standard". At this time, the system will be based on the error Error=T diff,actual -T diff,baseline =0.08℃ and its sign (positive value indicates high temperature), according to the preset fine-tuning rules and step size (such as wind speed increase δV=+0.01m / s, particle size decrease δD=-0.5μm), the current V final and D final Corrections are made, new drive instructions are generated, and the temperature difference is monitored again to form a closed-loop feedback control until the temperature difference reaches the baseline standard.

[0180] A smart farm environment control system based on big data analysis is used to execute the above-mentioned smart farm environment control method based on big data analysis. The system includes:

[0181] The data acquisition and fusion module is used to obtain the upper and lower surface temperature values of the leaves through infrared temperature sensors, the instantaneous stomatal conductance value through stomatal conductance sensors, and the air cavity water vapor pressure difference value through water vapor pressure sensors. These parameters are input into the multi-source data fusion algorithm to align the time series and remove noise, generating a dynamic physiological data set. The dynamic physiological data set is then passed to the temperature difference amplification discrimination module and the water vapor linkage analysis module.

[0182] The temperature difference increase discrimination module is used to perform difference calculation on the instantaneous temperature value of the upper surface of the leaf and the instantaneous temperature value of the lower surface of the leaf using a dynamic physiological data set, compare the temperature difference values of adjacent cycles, input the temperature difference sequence into the time series trend analysis to discriminate the temperature difference increase. If the temperature difference increase exceeds the preset threshold for three consecutive cycles, a transpiration activity enhancement signal is generated and transmitted to the wind speed and particle size control module;

[0183] The wind speed and particle size control module is used to increase the wind speed of the evaporative cooling unit by 0.5 m / s per cycle and decrease the water mist particle size by 5% per cycle through the transpiration activity enhancement signal, generate wind speed and particle size coordinated control instructions, and transmit the wind speed and particle size coordinated control instructions to the dynamic feedback execution module;

[0184] The water vapor linkage analysis module is used to perform Pearson correlation calculation on the instantaneous value of stomatal conductance and the air cavity water vapor pressure difference value through the dynamic physiological data set. If the absolute value of the correlation coefficient is greater than 0.8, the instantaneous value of stomatal conductance and the air cavity water vapor pressure difference value are input into the multivariate linear regression model to assign weight coefficients, generate water vapor linkage control coefficients, and pass the water vapor linkage control coefficients to the dynamic feedback execution module;

[0185] The dynamic feedback execution module is used to integrate the wind speed-particle size coordinated control instructions and the water vapor linkage control coefficient, and input the above parameters into the feedback control algorithm to dynamically map the control parameters of the evaporative cooling unit. When the water vapor linkage control coefficient is greater than 0.6, the wind speed is increased to 1.2 times the current value and the water mist particle size is reduced to 90% of the current value. Adjustments are continued until the temperature difference on the blade surface returns to the set baseline.

[0186] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A smart farm environment control method based on big data analysis, characterized in that: The following steps are involved: S1: The instantaneous temperature values of the upper and lower surfaces of the leaves are obtained through infrared temperature sensors, the instantaneous stomatal conductance value is obtained through stomatal conductance sensors, and the air cavity water vapor pressure difference value is obtained through water vapor pressure sensors. A multi-source data fusion algorithm is used to align the time series and remove noise to generate a dynamic physiological data set. S2: performing a difference calculation on the upper and lower surface temperatures of the leaf in the dynamic physiological data set, comparing the temperature difference values of adjacent cycles, and using time series trend analysis to determine the temperature difference increase. If the increase exceeds a threshold for three consecutive cycles, a transpiration activity enhancement signal is output; S3: Based on the transpiration activity enhancement signal, the wind speed of the evaporative cooling unit is adjusted to increase by 0.5 m / s per cycle and the water mist particle size is adjusted to decrease by 5% per cycle, thereby generating a wind speed-particle size coordinated control instruction; S4: Performing a Pearson correlation calculation on the stomatal conductance and the air cavity water vapor pressure difference in the dynamic physiological data set. If the absolute value of the correlation coefficient is greater than 0.8, a multivariate linear regression model is used to assign a weight coefficient to generate a water vapor linkage control coefficient.

2. The smart farm environment control method based on big data analysis according to claim 1 is characterized in that: The dynamic physiological data set specifically includes the instantaneous temperature value of the upper surface of the leaf, the instantaneous temperature value of the lower surface of the leaf, the instantaneous value of the stomatal conductance, and the air cavity water vapor pressure difference value. The transpiration activity enhancement signal includes the temperature difference increase, the number of consecutive cycles, and the temperature difference threshold. The wind speed-particle size coordinated control instruction includes the wind speed adjustment gradient and the particle size adjustment gradient. The water vapor linkage control coefficient specifically refers to the correlation coefficient threshold and the regression weight.

3. The smart farm environment control method based on big data analysis according to claim 2 is characterized in that: The multi-source data fusion algorithm includes a timestamp alignment algorithm, a sliding window standard deviation calculation method, a wavelet threshold denoising method and a weighted Kalman filter algorithm; The wind speed adjustment gradient and the particle size adjustment gradient are parameter solutions after optimizing the transpiration suppression efficiency based on a multiple regression model; The determination threshold of the absolute value of the correlation coefficient is the critical value verified by the t-test under the conditions of significance level α=0.01 and degree of freedom n-2; The feedback control algorithm is a dynamic parameter mapping rule based on proportional-integral-differential, with a proportional coefficient set to 0.8, an integral time constant to 10 seconds, and a differential time constant to 2 seconds.

4. The smart farm environment control method based on big data analysis according to claim 3 is characterized in that: The steps for acquiring the dynamic physiological data set are specifically as follows: S101: Collect the instantaneous values of the upper and lower surface temperatures of the blade output by the infrared temperature sensor, synchronously obtain the instantaneous conductance value output by the stomatal conductance sensor and the instantaneous value of the water vapor pressure difference output by the water vapor pressure sensor, and use a timestamp alignment algorithm to match the four types of parameters according to millisecond-level timestamps to generate synchronized time series parameters; The timestamp alignment algorithm uses the Coordinated Universal Time (UTC) time base to uniformly convert the timestamps of the sensor acquisition parameters into millisecond-level UTC time; S102: Based on the synchronization time series parameters, a sliding window standard deviation calculation method is used to remove outliers from the temperature value sequence, and a wavelet threshold denoising is performed on the stomatal conductance value sequence and the water vapor pressure difference value sequence, and the low-frequency features are retained after the high-frequency noise components are truncated to generate a reconstructed feature sequence; The wavelet threshold denoising adopts the Stein unbiased risk estimation method to determine the threshold and cut off the noise component with a frequency higher than 1kHz; S103: calling the reconstructed feature sequence, calculating the instantaneous value of the temperature difference between the upper and lower surfaces of the leaf, combining the instantaneous value of the stomatal conductance and the instantaneous value of the water vapor pressure difference, dynamically fusing the multi-dimensional parameters using a weighted Kalman filter algorithm, and assigning fusion weights based on the covariance matrix to generate a dynamic physiological data set; The fusion weights of the weighted Kalman filter algorithm are allocated according to the normalized values of the diagonal elements of the covariance matrix.

5. The smart farm environment control method based on big data analysis according to claim 4 is characterized in that: The steps for obtaining the transpiration activity enhancement signal are specifically as follows: S201: Obtaining upper and lower surface temperature data of each leaf in the dynamic physiological data set, aligning the upper and lower surface temperatures of the same leaf at the same time along the time axis using the sampling time as an index, performing a subtraction operation at each time point, calculating the temperature difference between the two surfaces of the leaf per unit time, and generating a temperature difference sequence corresponding to multiple time points; The unit time is the sensor sampling interval, which is set to 100 milliseconds; S202: Based on the temperature difference sequence, data segments of two adjacent cycles are intercepted according to the period division rule. For the temperature difference at each time point, the operation of subtracting the value of the previous cycle from the value of the next cycle is performed. The temperature difference changes at the corresponding time points between the cycles are calculated. All changes are arranged in chronological order to generate a temperature difference increase sequence for adjacent cycles. The cycle division rule is to define every 300 consecutive sampling points as a control cycle; S203: Call the adjacent period temperature difference amplification sequence, take three consecutive periods as the window unit, perform a summation operation on the amplification values of multiple time points in the window, compare the summation result with the preset transpiration activity amplification threshold item by item, and generate a transpiration activity enhancement signal when the cumulative amplification of any window exceeds the threshold. The transpiration activity increase threshold is a temperature difference cumulative increase range of 3.2°C to 4.5°C calibrated by the gradient descent method optimization experiment.

6. The smart farm environment control method based on big data analysis according to claim 5 is characterized in that: The steps for obtaining the wind speed-particle size coordinated control instruction are specifically as follows: S301: Based on the transpiration activity enhancement signal, the current wind speed value is called, and an incremental step is superimposed on the current value. The initial wind speed and a fixed step parameter are input into a periodic increasing function, and the superimposed value of the wind speed in each period is calculated to generate a wind speed increment value. The incremental step parameter is calibrated by the fluid mechanics model to calibrate the wind duct resistance coefficient of the evaporative cooling unit, and the step unit is m / s. 2 ; S302: Dividing nodes according to the period of the wind speed increment value, extracting the median value of the original water mist particle size distribution, reducing the upper and lower limits of the particle size range simultaneously in a geometrically decreasing manner, and compressing and translating the particle size distribution curve using a segmented iterative algorithm to generate a particle size decreasing gradient; The compression translation ratio of the segmented iterative algorithm is to reduce the particle size distribution interval width by 5% per cycle; S303: On the control cycle time axis, a Cartesian product mapping is performed on the numerical sequence of the wind speed increment value and the distribution parameter of the particle size decrement gradient to establish a two-dimensional parameter combination matrix of wind speed and particle size, and an instruction parameter set is generated through matrix coordinate transformation to obtain a wind speed and particle size coordinated control instruction; The matrix coordinate transformation rule is that the wind speed-particle size coupling coefficient is 0.7, and the linear correlation between wind speed and particle size is calibrated through experiments.

7. The smart farm environment control method based on big data analysis according to claim 6 is characterized in that: The steps for obtaining the water vapor linkage control coefficient are specifically as follows: S401: detecting the stomatal conductance and the air cavity water vapor pressure difference in the dynamic physiological data set, and calculating the covariance and standard deviation ratio between the stomatal conductance and the air cavity water vapor pressure difference based on the time series change value of the stomatal conductance and the time series change value of the air cavity water vapor pressure difference, to generate a Pearson correlation coefficient value; S402: calling the Pearson correlation coefficient value, comparing the absolute value of the correlation coefficient with a preset correlation threshold, and if the condition is met, establishing a linear combination equation for the stomatal conductance and the air cavity water vapor pressure difference by the least squares method to generate a weight distribution reference parameter; S403: Calling the weight distribution benchmark parameter, extracting the difference between the stomatal conductance at adjacent time points and the difference between the air cavity water vapor pressure difference at adjacent time points, integrating the ambient temperature sensor acquisition value and the leaf surface area measurement value, and using the formula: Calculate the ratio of the square root of the dynamic change difference to the weighted sum of the ambient temperature, and combine it with the regression model residual correction to generate the water vapor linkage control coefficient; Where W represents the water vapor linkage control coefficient, ΔG s is the difference in stomatal conductance at adjacent time points, ΔV p is the difference in water vapor pressure difference between adjacent time points in the air cavity, λ i is the ambient temperature sensor calibration coefficient, is the real-time measurement value of the temperature sensor, n is the number of sensors, r is the Pearson correlation coefficient, S leaf is the calibrated value of the blade surface area; The ambient temperature sensor calibration coefficient λ i The leaf surface area calibration value S is obtained by least square fitting the nonlinear relationship between ambient temperature and stomatal conductance. leaf is the mean value of the measured leaf projected area.

8. The smart farm environment control method based on big data analysis according to claim 7 is characterized in that: The method further comprises: S5: Integrate the wind speed-particle size coordinated control instruction and the water vapor linkage control coefficient, and dynamically map them to the evaporative cooling unit using a feedback control algorithm. When the water vapor linkage control coefficient is greater than 0.6, the wind speed is increased to 1.2 times the current value, and the water mist particle size is reduced to 90%. Continue to adjust until the blade surface temperature difference returns to the set baseline; The control parameters dynamically mapped by the feedback control algorithm include wind speed multiples and particle size percentages.

9. The smart farm environment control method based on big data analysis according to claim 8 is characterized in that: The steps of S5 are specifically as follows: S501: Integrate the real-time wind speed value and the water mist particle size parameter in the wind speed-particle size coordinated control instruction, perform a proportional integral operation on the wind speed value and the particle size parameter, use the water vapor linkage control coefficient as a dynamic weight factor, construct a linear combination relationship between wind speed and particle size, perform a boundary check on the calculation result and the input parameter range of the evaporative cooling unit, and generate a coordinated control parameter; The integral time constant of the proportional integral operation is set to 5 seconds, and the boundary check range is wind speed 0-15m / s and particle size 10-100μm; S502: calling the water vapor linkage control coefficient in the coordinated control parameter, using a threshold comparator to compare the coefficient with a preset threshold, and when the coefficient exceeds the threshold, extracting the dynamic adjustment coefficient of the current wind speed value and the proportional adjustment parameter of the particle size value, performing a multiplication operation of the wind speed value and a proportional reduction operation of the particle size value, and generating a control activation signal; The hysteresis interval of the threshold comparator is set to ±0.05 to avoid frequent switching caused by parameter fluctuations; S503: Based on the actuator interface of the evaporative cooling unit, the target wind speed and target particle size in the control activation signal are converted into device drive instructions, the real-time value of the blade surface temperature difference sensor is synchronously collected, and the absolute difference between the temperature difference value and the baseline value is calculated. When the difference does not meet the preset convergence standard, the wind speed and particle size outputs are cyclically corrected according to the adjustment step size to generate a temperature difference baseline compliance status; The preset convergence standard is that the absolute temperature difference is less than 0.5°C, the adjustment step is the wind speed change ±0.2m / s, and the particle size change ±1μm, and closed-loop convergence is achieved through PID control.

10. The smart farm environment control system based on big data analysis is characterized by: The system is used to implement the smart farm environment control method based on big data analysis according to any one of claims 1 to 9, and the system includes: The data acquisition and fusion module is used to obtain the upper surface temperature value and the instantaneous temperature value of the lower surface of the blade through the infrared temperature sensor, obtain the instantaneous value of the stomatal conductance through the stomatal conductance sensor, and obtain the air cavity water vapor pressure difference value through the water vapor pressure sensor. The above parameters are input into the multi-source data fusion algorithm to align the time series and remove noise, generate a dynamic physiological data set, and transmit the dynamic physiological data set to the temperature difference increase determination module and the water vapor linkage analysis module; The temperature difference increase determination module is used to perform difference calculation on the instantaneous temperature value of the upper surface of the leaf and the instantaneous temperature value of the lower surface of the leaf using a dynamic physiological data set, compare the temperature difference values of adjacent cycles, input the temperature difference sequence into the time series trend analysis to determine the temperature difference increase. If the temperature difference increase exceeds a preset threshold for three consecutive cycles, a transpiration activity enhancement signal is generated and transmitted to the wind speed and particle size control module; a wind speed and particle size control module, configured to increase the wind speed of the evaporative cooling unit by 0.5 m / s per cycle and decrease the water mist particle size by 5% per cycle using the transpiration activity enhancement signal, generate a wind speed-particle size coordinated control instruction, and transmit the wind speed-particle size coordinated control instruction to the dynamic feedback execution module; a water vapor linkage analysis module, configured to perform a Pearson correlation calculation on the instantaneous value of stomatal conductance and the air cavity water vapor pressure difference value using the dynamic physiological data set; if the absolute value of the correlation coefficient is greater than 0.8, input the instantaneous value of stomatal conductance and the air cavity water vapor pressure difference value into a multivariate linear regression model to assign a weight coefficient, generate a water vapor linkage control coefficient, and transmit the water vapor linkage control coefficient to the dynamic feedback execution module; A dynamic feedback execution module is used to integrate the wind speed-particle size coordinated control instruction and the water vapor linkage control coefficient, and input the above parameters into the feedback control algorithm to dynamically map the control parameters of the evaporative cooling unit. When the water vapor linkage control coefficient is greater than 0.6, the wind speed is increased to 1.2 times the current value and the water mist particle size is reduced to 90% of the current value. Adjustments are continuously made until the temperature difference on the blade surface returns to the set baseline.

Images