A forage crop transpiration real-time monitoring system

Abstract

The present application relates to the technical field of agricultural monitoring, and specifically discloses a real-time monitoring system for transpiration of forage crops. The system comprises a multi-source data sensing module, a stress response correction module, a dynamic self-adaptive calibration module and a transpiration fusion calculation module. By collecting physiological, micro-meteorological and soil data of crops in real time, the water stress state is identified and a stress correction factor is generated. The initial transpiration estimation value is dynamically calibrated in combination with a meteorological mutation compensation mechanism, and finally a high-precision real-time transpiration monitoring value is output through neural network fusion calculation. The system improves the adaptability to the water condition of crops and environmental changes, and provides reliable data support for precision irrigation.

Description

Technical Field

[0001] This invention belongs to the field of agricultural monitoring technology, specifically relating to a real-time monitoring system for the transpiration of forage crops. Background Technology

[0002] In the fields of precision agriculture and smart irrigation, accurate monitoring of crop water use efficiency is a key technological link in optimizing water resource management and increasing crop yield. As a core physiological indicator reflecting crop water status, accurate acquisition of transpiration is of great significance for achieving on-demand irrigation and avoiding water waste.

[0003] Real-time monitoring of transpiration from forage crops is an important branch of agricultural hydrological monitoring. This technology aims to provide direct data support for precise irrigation decisions for forage crops by measuring the amount of water consumed by the crops' own transpiration, thereby ensuring the quality and yield of forage.

[0004] Existing technologies typically employ the evapotranspiration-evaporation difference method for estimation, but this method has limitations: it does not fully consider the water stress faced by forage crops at different growth stages, nor does it effectively address sudden changes in field weather conditions. These interfering factors lead to systematic biases in the final transpiration calculation results; for example, the error can be as high as 25% under drought conditions. Furthermore, traditional measurement schemes generally lack dynamic calibration mechanisms, making it difficult to adapt to the complex and nonlinear changes in the field environment, resulting in insufficient reliability and universality of the measurement results.

[0005] Therefore, how to achieve high-precision, adaptive, real-time monitoring of transpiration of forage crops in complex field environments has become a technical challenge that urgently needs to be solved in this field. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art in estimating the transpiration of forage crops by evapotranspiration difference method, which is due to the neglect of the influence of crop water stress and sudden changes in field meteorological conditions, resulting in large calculation errors and lack of dynamic calibration capability. The present invention provides a real-time monitoring system for the transpiration of forage crops.

[0007] To achieve the above objectives, this invention provides a real-time monitoring system for transpiration of forage crops. A multi-source data sensing module is used to collect multi-dimensional field data directly related to the crop transpiration process in real time, including crop physiological data, field micrometeorological data, and soil environmental data.

[0008] The stress response correction module is connected to the multi-source data sensing module to identify the crop's water stress status based on real-time collected crop physiological data and generate a dynamic stress correction factor.

[0009] The dynamic adaptive calibration module is connected to both the multi-source data sensing module and the stress response correction module. It is used to calibrate the initial evapotranspiration estimate based on the traditional model in real time based on the abrupt change characteristics of field micrometeorological data and the stress correction factor, so as to generate a corrected evapotranspiration baseline value.

[0010] The evapotranspiration fusion calculation module is further connected to the dynamic adaptive calibration module and the stress response correction module. It is used to perform nonlinear fusion calculations on the corrected evapotranspiration baseline value and the stress correction factor, and finally output a high-precision real-time evapotranspiration monitoring value.

[0011] The multi-source data sensing module specifically includes a canopy infrared temperature sensor, a stem flow meter, an air temperature and humidity sensor, a wind speed and direction sensor, a solar radiation sensor, and a soil moisture sensor.

[0012] The canopy infrared temperature sensor measures the temperature distribution on the surface of the crop canopy in a non-contact manner at a sampling frequency of 1 time / min.

[0013] The stem flow meter uses the principle of thermal equilibrium to directly measure the sap flow rate of crop stems at a sampling frequency of 30 seconds per measurement.

[0014] The air temperature and humidity sensor, together with the wind speed and direction sensor, constitutes the field micro-meteorological measurement unit, with a sampling frequency of 10 seconds per sampling.

[0015] Solar radiation sensors continuously monitor the total radiation intensity reaching the crop canopy.

[0016] Soil moisture sensors are deployed within a depth of 0cm to 40cm in the main root layer of the crop, measuring soil volumetric water content at a frequency of 5 minutes. All of the above sensors aggregate data through IoT nodes and transmit encrypted data packets wirelessly to the system's central processing unit.

[0017] The internal processing flow of the stress response correction module is as follows:

[0018] First, the canopy infrared temperature measurement value and the air temperature and humidity measurement value are obtained in real time from the multi-source data sensing module. The canopy temperature difference is obtained by calculating the difference between the crop canopy temperature and the air temperature.

[0019] At the same time, the module receives crop stem sap flow rate data measured by the stem flow meter.

[0020] Set a preset crown temperature difference threshold range and stem flow rate benchmark range based on crop type and growth stage.

[0021] When the real-time crown temperature difference exceeds the upper limit of the preset threshold range for 3 minutes and the real-time stem flow rate is lower than the lower limit of the benchmark range, the module determines that the crop has entered a state of water stress.

[0022] At this point, the module initiates the stress correction factor calculation program. This program is based on the weighted product of the deviation of the crown-temperature difference and the stem flow rate attenuation rate, and introduces a stress intensity coefficient that accumulates over time. Finally, it outputs a dynamic stress correction factor ranging from 0.5 to 1.8. If no water stress is detected, the stress correction factor defaults to a value of 1, indicating no stress effect.

[0023] The operating mechanism of the dynamic adaptive calibration module is as follows:

[0024] This module has a built-in initial evapotranspiration estimation model based on the Penman Montes formula, which uses real-time field micrometeorological data as the main input.

[0025] The module continuously monitors the data streams from the wind speed and direction sensors and the air temperature and humidity sensors, and calculates the rate of change of meteorological parameters during adjacent sampling weeks.

[0026] When the absolute value of the rate of change of any meteorological parameter exceeds its respective preset mutation threshold within 1 minute, the module immediately determines that a sudden change in meteorological conditions has occurred in the field and triggers the calibration procedure.

[0027] The calibration procedure first calculates a meteorological change compensation coefficient based on the direction and magnitude of the meteorological change, with a value ranging from 0.8 to 1.5.

[0028] The module then reads the stress correction factor provided in real time by the stress response correction module.

[0029] The dynamic adaptive calibration module multiplies the initial transpiration estimate, the meteorological abrupt change compensation coefficient, and the stress correction factor. The result is the corrected baseline transpiration value. This baseline value serves as the basis for subsequent fusion calculations, eliminating the systematic errors jointly introduced by meteorological abrupt changes and crop stress.

[0030] The transpiration fusion calculation module performs the final high-precision transpiration calculation. This module uses a two-layer neural network structure for nonlinear fusion.

[0031] The first layer of the neural network is fed into the transpiration baseline value output by the dynamic adaptive calibration module and the stress correction factor output by the stress response correction module.

[0032] The second layer of the neural network additionally introduces real-time soil moisture data provided by the multi-source data sensing module as an environmental constraint.

[0033] The neural network was trained and validated using a dataset of measured transpiration rates of forage crops under different water treatments during two historical growing seasons. The activation function of the hidden layer was modified linear unit, and the output layer was linear activation function.

[0034] The training objective is to minimize the root mean square error between the neural network output value and the actual transpiration measured synchronously using a weighing lysimeter.

[0035] The trained neural network can capture the complex nonlinear relationship between the evapotranspiration baseline under stress and soil moisture, thereby finely adjusting the evapotranspiration baseline and finally outputting a real-time evapotranspiration monitoring value in mm / h. The refresh frequency of this output value is consistent with the data update frequency of the multi-source data sensing module, which is 1 time / min.

[0036] As one embodiment of the present invention, the system also integrates a data post-processing and quality control unit, which is connected after the evapotranspiration fusion calculation module.

[0037] The data post-processing and quality control unit performs real-time analysis on the real-time evapotranspiration monitoring value sequence output by the fusion computing module, and applies the Laida criterion based on a sliding time window to detect and remove outliers.

[0038] For outlier data points that were removed, linear interpolation of the two adjacent valid data points was used to fill in the gaps, ensuring the continuity and rationality of the output data sequence. Simultaneously, this unit also calculates the average and standard deviation of transpiration over every 10-minute timescale, and stores these statistics along with the original monitoring values ​​in the system database, providing comprehensive data support for long-term irrigation decisions and water use efficiency analysis.

[0039] In another embodiment of the present invention, the central processing unit of the system is also equipped with a remote communication interface, supporting 4G or 5G mobile communication networks to upload the final evapotranspiration monitoring results and related stress alarm information to the cloud management platform in real time. Users can access the cloud platform through authorized terminal devices to view the field evapotranspiration dynamics in real time and receive irrigation suggestion instructions automatically generated by the system based on the cumulative evapotranspiration value exceeding the preset irrigation threshold.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] 1. This invention fundamentally solves the inherent defect of the traditional evapotranspiration difference method in being insensitive to crop water stress and sudden changes in field weather by constructing a system architecture that combines multi-source data sensing stress response correction dynamic adaptive calibration with evapotranspiration fusion calculation.

[0042] 2. The stress response correction module uses the dual criteria of crown temperature difference and stem flow rate to accurately identify and quantify the crop's water stress state, and transforms this physiological information into a calculable stress correction factor. The crop's own physiological response is directly embedded into the transpiration calculation model.

[0043] 3. The dynamic adaptive calibration module improves the system's robustness to short-term and drastic changes in the field environment by capturing sudden changes in meteorological parameters in real time and introducing a meteorological change compensation coefficient, thereby suppressing estimation bias caused by such interference.

[0044] 4. The final evapotranspiration fusion calculation module utilizes a trained neural network model to achieve nonlinear deep fusion of calibrated baseline evapotranspiration, crop stress factors, and soil environmental constraints. This process fully simulates the complexity of the field evapotranspiration process, thereby outputting real-time evapotranspiration monitoring values ​​with high reliability and strong adaptability.

[0045] 5. The entire system forms a complete technical closed loop from data perception to stress identification, dynamic calibration, and intelligent integration, providing a solid data foundation and decision-making basis for the precise irrigation management of forage crops and the efficient utilization of water resources in field environments. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the overall technical architecture of the real-time monitoring system for transpiration of forage crops proposed in this invention;

[0047] Figure 2 This is a schematic diagram of the core principle framework of the stress response correction module in this invention;

[0048] Figure 3 This is a flowchart of the dynamic adaptive calibration module in this invention.

[0049] Figure 4 This is a schematic diagram of the two-layer neural network structure of the transpiration fusion calculation module in this invention;

[0050] Figure 5 This is a schematic diagram of the multi-level interaction relationship and data flow between the multi-source data sensing module and various functional modules in this invention; Detailed Implementation

[0051] This embodiment details the specific implementation structure and operating mechanism of a real-time monitoring system for transpiration of forage crops. Please refer to the appendix. Figure 1 , attached Figure 1 This is a schematic diagram of the overall technical architecture of the real-time monitoring system for transpiration of forage crops proposed in this invention. The system consists of a multi-source data sensing module, a stress response correction module, a dynamic adaptive calibration module, a transpiration fusion calculation module, a data post-processing and quality control unit, and a remote communication interface. Each module achieves data interaction and command coordination through a central processing unit, forming a complete closed-loop monitoring system.

[0052] Please refer to the appendix. Figure 5 , attached Figure 5This is a schematic diagram of the multi-level interaction relationship and data flow between the multi-source data sensing module and various functional modules in this invention. The multi-source data sensing module is responsible for collecting multi-dimensional field data related to crop transpiration, including crop physiological data, field micro-meteorological data and soil environmental data.

[0053] The multi-source data sensing module consists of a canopy infrared temperature sensor, stem flow meter, air temperature and humidity sensor, wind speed and direction sensor, solar radiation sensor, and soil moisture sensor.

[0054] The canopy infrared temperature sensor is installed 1.5 mm / h above the crop canopy and measures the surface temperature distribution of the canopy non-contactly at a frequency of 1 time / min. Its measurement accuracy is ±0.2℃. The data is transmitted to the central processing unit wirelessly through an Internet of Things node.

[0055] The stem flow meter is based on the principle of heat balance and directly monitors the sap flow rate of crop stems at a sampling frequency of 30 seconds per sample. The sensor probe is in close contact with the stem epidermis to ensure that the heat flux measurement error is less than 5%.

[0056] An air temperature and humidity sensor, together with a wind speed and direction sensor, constitutes a field micro-meteorological measurement unit. The unit is installed at a height of 0.5m above the top of the crop canopy, and the sampling frequency is 10s per sampling.

[0057] The air temperature and humidity sensors measure temperatures from -20℃ to 60℃ and relative humidity from 0% to 100%, respectively, while the wind speed and direction sensors measure wind speeds from 0m / s to 30m / s and azimuths from 0° to 360°.

[0058] A solar radiation sensor is horizontally deployed at the top of the canopy to continuously monitor total radiation intensity, with a spectral response range of 300 nm to 3000 nm and a sensitivity of 10 μV / W / m². A soil moisture sensor is deployed in the main root zone of the crop at a depth of 0 cm to 40 cm, measuring soil volumetric water content at a frequency of 5 minutes per measurement, with a measurement accuracy of ±2%.

[0059] All sensor data is encrypted and packaged by IoT nodes using a 128-bit advanced encryption standard algorithm and transmitted to the data buffer of the central processing unit via a wireless communication protocol.

[0060] The stress response correction module is connected to the multi-source data sensing module to identify crop water stress status and generate dynamic stress correction factors.

[0061] Please refer to the attached document. Figure 2 , attached Figure 2This is a schematic diagram of the core principle framework of the stress response correction module in this invention. The module first acquires the canopy temperature data measured by the canopy infrared temperature sensor and the air temperature data measured by the air temperature and humidity sensor in real time, and calculates the difference between the canopy temperature and the air temperature, i.e., the canopy temperature difference.

[0062] At the same time, the module receives data on the sap flow rate of crop stems measured by the stem flow meter.

[0063] The module has preset crown temperature difference threshold range and stem flow rate benchmark range based on crop type and growth stage. For example, the crown temperature difference threshold range for alfalfa during the flowering period is -2℃ to 3℃, and the stem flow rate benchmark range is 20g / h to 60g / h.

[0064] When the real-time crown temperature difference exceeds the upper limit of the preset threshold range for 3 minutes and the real-time stem flow rate is lower than the lower limit of the benchmark range, the module determines that the crop has entered a state of water stress.

[0065] The stress correction factor calculation program is based on the weighted product of the degree of deviation of crown temperature difference and stem flow rate attenuation rate, and introduces the stress intensity coefficient accumulated over time.

[0066] Specifically, the deviation of the canopy temperature difference is defined as the absolute value of the difference between the real-time canopy temperature difference and the upper limit of the threshold interval, and the stem flow rate attenuation rate is defined as the absolute value of the ratio of the real-time stem flow rate to the lower limit of the benchmark interval minus 1. The weighting weights are determined to be 0.6 and 0.4 based on crop sensitivity experiments.

[0067] The stress intensity coefficient accumulates at a rate of 0.05 per minute, with a maximum value of 1.2.

[0068] The final output stress correction factor is strictly controlled within the range of 0.5 to 1.8. If no water stress is detected, it is maintained at 1 by default. This factor is transmitted in real time to the dynamic adaptive calibration module and the transpiration fusion calculation module via the data bus of the central processing unit.

[0069] The dynamic adaptive calibration module connects to both the multi-source data sensing module and the stress response correction module to perform real-time calibration of the initial evapotranspiration estimate.

[0070] Please refer to the appendix. Figure 3 , attached Figure 3 This is a logical flowchart of the dynamic adaptive calibration module in this invention. The module incorporates an initial evapotranspiration estimation model based on the Penman Montes formula. Its inputs are real-time field micro-meteorological data, including air temperature, relative humidity, wind speed, and solar radiation intensity. The module continuously monitors the data streams from wind speed and direction sensors and air temperature and humidity sensors, calculating the rate of change of meteorological parameters during adjacent sampling periods.

[0071] The rate of change is defined as the difference between the current sample value and the previous sample value divided by the previous sample value.

[0072] When the absolute value of the rate of change of any meteorological parameter exceeds the preset abrupt change threshold within 1 minute, the module determines that a sudden change in meteorological conditions has occurred.

[0073] The threshold for sudden change is set based on statistical analysis of historical meteorological data. For example, the threshold for wind speed change rate is 30%, and the threshold for air temperature change rate is 20%.

[0074] After the calibration procedure is triggered, the module first calculates the meteorological change compensation coefficient based on the direction and magnitude of the meteorological change.

[0075] If meteorological parameters change in a direction that intensifies evaporation, the compensation coefficient is taken as 1.0 to 1.5; if they change in a direction that inhibits evaporation, it is taken as 0.8 to 1.0.

[0076] The specific calculation uses linear interpolation, with the ratio of the rate of change to the threshold as the interpolation factor.

[0077] Subsequently, the module reads the stress correction factor provided in real time by the stress response correction module.

[0078] The dynamic adaptive calibration module multiplies the initial evapotranspiration estimate, the meteorological abrupt change compensation coefficient, and the stress correction factor to generate a corrected baseline evapotranspiration value.

[0079] This baseline value is transmitted to the evapotranspiration fusion calculation module via a data interface, eliminating the systematic errors introduced by both meteorological abrupt changes and crop stress.

[0080] The transpiration fusion calculation module connects the dynamic adaptive calibration module and the stress response correction module to perform the final high-precision transpiration calculation.

[0081] Please refer to the attached document. Figure 4 , Figure 4 This is a schematic diagram of the two-layer neural network structure of the transpiration fusion calculation module in this invention. This module uses a two-layer neural network structure for nonlinear fusion.

[0082] The first layer of the neural network is fed into the transpiration baseline value output by the dynamic adaptive calibration module and the stress correction factor output by the stress response correction module.

[0083] The second layer additionally introduces real-time soil moisture data provided by a multi-source data sensing module as an environmental constraint.

[0084] The neural network structure includes an input layer, a hidden layer, and an output layer. The input layer has 3 nodes, the hidden layer has 8 nodes, and the output layer has 1 node.

[0085] The hidden layer activation function uses the modified linear unit, and its mathematical expression is: ;

[0086] The output layer employs a linear activation function. The neural network is trained using a dataset of measured transpiration rates of forage crops under different water treatments from two historical growing seasons. The training dataset contains 5000 samples, each including a baseline transpiration rate, a stress correction factor, soil volumetric water content, and the actual transpiration rate measured by a gravimetric lysimeter. The training process uses an adaptive moment estimation algorithm with an initial learning rate of 0.001, a batch size of 32, and 200 training epochs.

[0087] The training objective was to minimize the root mean square error (RMSE) between the neural network output and the actual transpiration rate, with the final validation set error below 0.05 mm / h. The trained neural network was able to capture the complex nonlinear relationship between the baseline transpiration rate, stress conditions, and soil moisture, enabling fine-tuning of the baseline transpiration rate. The final output was a real-time transpiration rate monitoring value in mm / h, refreshed at a frequency consistent with the multi-source data sensing module's update frequency of 1 time / min. The output data was transmitted to the data post-processing and quality control unit via the central processing unit's shared memory area.

[0088] The data post-processing and quality control unit is connected after the evapotranspiration fusion calculation module to perform real-time analysis and quality control on the real-time evapotranspiration monitoring value sequence.

[0089] This unit uses the Raida criterion based on a sliding time window to detect outliers, with a sliding window length of 10 data points.

[0090] For each data sequence within a window, the cell calculates its mean and standard deviation.

[0091] If the absolute value of the deviation of a data point from the mean exceeds three times the standard deviation, it is considered an outlier and is removed.

[0092] For the outlier data points that were removed, the cell is filled by linear interpolation of the two adjacent valid data points.

[0093] The interpolation formula is: ;

[0094] in To fill in the value, and These are valid data points before and after the event. Simultaneously, this unit calculates the average and standard deviation of evapotranspiration over every 10-minute timescale, and stores these statistics along with the original monitoring values ​​in the system database.

[0095] The database uses a relational structure, and each record includes a timestamp, evapotranspiration monitoring value, 10-minute average, 10-minute standard deviation, and data quality flags. Stored procedures employ transaction mechanisms to ensure data atomicity and consistency.

[0096] The system's central processing unit is equipped with a remote communication interface, supporting 4G or 5G mobile communication networks.

[0097] This interface uses an embedded modem module to establish a connection with the mobile network base station via a user identification card. The communication protocol is Transmission Control Protocol (TCP) and Internet Protocol (IP) suite, and the data is encapsulated in JavaScript object representation format.

[0098] The remote communication interface uploads the final evapotranspiration monitoring results and related stress alarm information to the cloud management platform in real time, at a frequency of once per minute. The stress alarm information includes the duration and intensity level of stress correction factors below 0.7 or above 1.5.

[0099] Users can access the cloud platform through authorized terminal devices to view field evapotranspiration dynamics in real time. The cloud platform also provides data visualization services, including evapotranspiration time series curves, stress state distribution maps, and soil moisture thermograms.

[0100] The system automatically generates irrigation suggestions based on the cumulative transpiration exceeding a preset irrigation threshold. The irrigation threshold is dynamically adjusted according to the crop water requirement model. For example, the daily transpiration threshold for alfalfa during its vegetative growth period is 5 mm.

[0101] Irrigation recommendations are pushed to users' mobile devices via the cloud platform or sent directly to the irrigation control system.

[0102] Please refer to the attached document for comprehensive information. Figure 1 With appendix Figure 5 The system's central processing unit uses a multi-core microprocessor with a clock speed of 1.8GHz, 4GB of memory, and 128GB of storage.

[0103] After the system is powered on, it first initializes each sensor and communication interface, and loads the neural network model parameters and configuration parameters.

[0104] The multi-source data sensing module collects data at a preset frequency and transmits encrypted data packets to the data receiving queue of the central processing unit via a wireless communication protocol.

[0105] The stress response correction module extracts canopy temperature, air temperature and stem flow rate data from the queue and performs stress status judgment and correction factor calculation.

[0106] The dynamic adaptive calibration module synchronously extracts micrometeorological data from the queue, runs the initial evapotranspiration estimation model, and generates a baseline value for evapotranspiration by combining meteorological abrupt change detection and stress correction factors.

[0107] The evapotranspiration fusion calculation module receives baseline values, stress correction factors, and soil moisture data, and calculates the final evapotranspiration monitoring value through a neural network.

[0108] The data post-processing and quality control unit performs anomaly processing and statistical calculations on the monitored value sequence, stores the results in the database, and triggers the remote communication interface for uploading.

[0109] The entire system operates at a rate of 1 time per minute. The central processing unit uses a hardware timer for precise scheduling to ensure the timing synchronization of each module.

[0110] The system also has a watchdog mechanism that automatically restarts the relevant processes and records the fault log when any module times out or abnormal data accumulates.

[0111] This embodiment provides an alternative implementation scheme for a real-time monitoring system of forage crop transpiration, focusing on its adaptive enhancement mechanism and multimodal data fusion strategy under extreme weather conditions.

[0112] Please refer to the attached document. Figure 1 With appendix Figure 3 In this embodiment, the dynamic adaptive calibration module introduces a multi-parameter collaborative analysis mechanism in the determination of meteorological sudden changes.

[0113] The module not only calculates the rate of change of individual meteorological parameters, but also analyzes the coupling relationship between parameters simultaneously, such as the synergistic abrupt change index of wind speed and air temperature.

[0114] The co-mutation index is defined as the absolute value of the product of the rate of change of wind speed and the rate of change of air temperature. When the index exceeds 0.1, the module determines it to be a strong co-mutation event.

[0115] At this point, a nonlinear adjustment function is introduced into the calculation of the meteorological abrupt change compensation coefficient, and its value range is extended to 0.7 to 1.6, so as to more accurately reflect the superimposed impact of complex meteorological abrupt changes on evapotranspiration.

[0116] In this embodiment, the stress response correction module adds a soil moisture trend factor as an auxiliary input for stress criteria.

[0117] Please refer to the appendix. Figure 2 The module analyzes the soil moisture content data from 0cm to 40cm depth provided by the soil moisture sensor in real time and calculates the slope of its 5-minute moving average.

[0118] When the slope is consistently negative and the absolute value exceeds 0.5% / min, the module provides an early warning of potential moisture stress risk and adjusts the lower limit of the crown temperature difference threshold range to a more sensitive level, such as from -2℃ to -1℃.

[0119] This mechanism enables the system to identify stress trends in advance when soil moisture continues to decline but before the stem flow rate threshold is triggered, thereby improving the response speed and accuracy of stress correction factors.

[0120] In this embodiment, the transpiration fusion calculation module employs a triple neural network structure to further optimize the nonlinear fusion effect. Please refer to the appendix. Figure 4 The first layer of the neural network takes as input a baseline value for transpiration, a stress correction factor, and soil moisture data. The second layer incorporates solar radiation intensity and relative humidity as environmental modulation factors. The third layer outputs the final transpiration monitoring value. The neural network training dataset has been expanded to 8,000 samples, covering meteorological scenarios such as extreme drought, high temperature, and strong wind.

[0121] During training, a regularization term is introduced into the loss function to prevent overfitting, and the regularization coefficient is set to 0.01.

[0122] In this embodiment, the data post-processing and quality control unit integrates a dynamic threshold adjustment function.

[0123] The unit monitors the ambient background noise level in real time and dynamically adjusts the standard deviation multiple of the Laida criterion based on changes in wind speed and solar radiation intensity.

[0124] For example, when the wind speed exceeds 5 m / s, the standard deviation factor is adjusted from 3 to 2.5 to enhance the sensitivity of outlier detection.

[0125] Meanwhile, the unit introduces a data credibility scoring mechanism, with scores based on sensor data integrity, neural network output confidence, and environmental consistency verification results. Data points with scores below 0.8 will be marked as low credibility and highlighted in a special color on the cloud platform to remind users to conduct manual review.

[0126] In this embodiment, the remote communication interface supports multi-network redundancy switching. When the 4G or 5G signal strength is below -90dBm, the interface automatically switches to the satellite communication backup link to ensure uninterrupted data upload.

[0127] The cloud management platform adds a multi-dimensional data correlation analysis function, which can simultaneously display the spatiotemporal distribution map of evapotranspiration monitoring values, meteorological data, soil moisture and stress alarms, and supports the generation of irrigation efficiency reports on a daily, weekly and monthly basis.

[0128] The system also integrates third-party weather forecast data, uses machine learning models to predict evapotranspiration trends for the next 6 hours, and generates irrigation recommendations in advance.

[0129] This embodiment focuses on describing the system's hardware deployment scheme and energy management strategy to ensure its long-term stable operation in field environments. The sensor nodes of the multi-source data sensing module utilize a solar power system. Solar panels are mounted on top of the sensor bracket, with the tilt angle adjusted according to the local latitude to maximize light energy collection efficiency. The charge / discharge controller features maximum power point tracking to ensure effective charging even in low-light conditions. The battery pack has a 7-day battery life, supporting uninterrupted operation during continuous cloudy or rainy weather.

[0130] The central processing unit is housed in a waterproof enclosure with a rating of 65, and operates within a temperature range of -30°C to 70°C. The enclosure is equipped with a temperature control module that activates fan cooling when the internal temperature exceeds 40°C and heat sinks to maintain warmth when the temperature drops below -10°C.

[0131] The system adopts a low-power design, with a standby power consumption of 2W and a peak power consumption of 8W. The modem module of the remote communication interface supports sleep mode, automatically entering a low-power state when there is no data transmission, thus reducing overall energy consumption.

[0132] In this embodiment, the sensor data acquisition link incorporates local caching and a resume mechanism. Each IoT node is equipped with a 4GB storage card; when wireless communication is interrupted, the acquired data is temporarily stored locally. After communication is restored, the node automatically retrieves the unuploaded data packets and retransmits them in chronological order. The central processing unit's data receiving queue employs a priority scheduling algorithm to ensure that real-time data is processed first, while historical data is retransmitted asynchronously.

[0133] The system installation and calibration process includes sensor location optimization, local fine-tuning of the neural network model, and communication link testing.

[0134] The installation height and angle of the canopy infrared temperature sensor are dynamically adjusted according to the crop height to ensure a canopy coverage of more than 90%. The stem flow meter probe is pre-calibrated before installation, and its measurement error is verified to be less than 3% using a standard liquid flow simulation device.

[0135] After deployment, the neural network model was fine-tuned using 100 locally collected samples to adapt to the soil and crop characteristics of specific fields.

[0136] Communication link testing includes signal strength mapping and data transmission packet loss rate statistics to ensure that the cloud platform access latency is less than 2 seconds.

[0137] The system maintenance mechanism includes periodic sensor calibration, battery status monitoring, and remote software upgrades. Sensor calibration is performed every three months, using standard reference equipment for cross-validation. Battery status is monitored via voltage and internal resistance; the system automatically sends a replacement alarm when the battery capacity decays to 80% of its initial value. Software upgrade packages are distributed via a cloud platform using differential upgrade technology to reduce data transmission volume, and the upgrade process requires no manual intervention.

[0138] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0139] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A system for real-time monitoring of transpiration of forage crops, characterized in that, The method comprises the following steps: a multi-source data sensing module for real-time acquisition of multi-dimensional field data directly related to crop transpiration process, including crop physiological data, field microclimate data and soil environmental data; a stress response correction module connected to the multi-source data sensing module for identifying the water stress state of crops based on real-time acquisition of crop physiological data and generating a dynamic stress correction factor; the internal processing flow of the stress response correction module is as follows: firstly, the crown layer infrared temperature measurement value and the air temperature and humidity measurement value are obtained from the multi-source data sensing module in real time, and the crown-air temperature difference is obtained by calculating the difference between the crop crown layer temperature and the air temperature; at the same time, the stem flow meter measured crop stem sap flow rate data is received; the crown-air temperature difference threshold interval and the stem flow rate reference interval based on the preset crop species and growth stage are set; when the real-time crown-air temperature difference continuously exceeds the upper limit of the preset threshold interval and the real-time stem flow rate is lower than the lower limit of the reference interval, the module determines that the crop enters the water stress state, and starts the stress correction factor calculation program; the stress correction factor calculation program is based on the weighted product of the crown-air temperature difference deviation and the stem flow rate attenuation rate, and introduces the stress intensity coefficient accumulated over time; the stress intensity coefficient is accumulated at a rate of 0.05 per minute, and the maximum value is 1.2; the final output stress correction factor range is controlled between 0.5 and 1.8; a dynamic adaptive calibration module connected to the multi-source data sensing module and the stress response correction module for real-time calibration of the initial transpiration amount estimated value based on the traditional model according to the mutation characteristics of the field microclimate data and the stress correction factor to generate a corrected transpiration amount reference value; a transpiration amount fusion calculation module further connected to the dynamic adaptive calibration module and the stress response correction module for nonlinear fusion calculation of the corrected transpiration amount reference value and the stress correction factor to finally output a high-precision real-time transpiration amount monitoring value, and the transpiration amount fusion calculation module adopts a double-layer neural network structure for nonlinear fusion.

2. The system according to claim 1, wherein the system is characterized by: The multi-source data sensing module specifically includes a crown layer infrared temperature sensor, a stem flow meter, an air temperature and humidity sensor, a wind speed and direction sensor, a solar radiation sensor and a soil moisture sensor; The crown layer infrared temperature sensor measures the temperature distribution of the crop crown layer surface in a non-contact manner at a fixed sampling frequency; The stem flow meter directly measures the sap flow rate of the crop stem at a fixed sampling frequency based on the heat balance principle; The air temperature and humidity sensor and the wind speed and direction sensor together constitute a field microclimate measurement unit; The solar radiation sensor continuously monitors the total radiation intensity reaching the crop canopy.

3. The system according to claim 1, wherein the system is characterized by: The dynamic adaptive calibration module has an initial transpiration amount estimation model based on the Penman-Monteith formula, which takes the real-time acquisition of field microclimate data as the main input quantity; The dynamic adaptive calibration module continuously monitors the data flow of the wind speed and direction sensor and the air temperature and humidity sensor, and calculates the change rate of meteorological parameters between adjacent sampling periods; The dynamic self-adaptive calibration module multiplies the initial transpiration estimation value, the meteorological mutation compensation coefficient and the stress correction factor to obtain a corrected transpiration reference value.

4. The system according to claim 1, wherein the system is characterized by: The first layer of the double-layer neural network takes the transpiration reference value output by the dynamic self-adaptive calibration module and the stress correction factor output by the stress response correction module as inputs. The second layer of the double-layer neural network takes the real-time soil moisture data provided by the multi-source data sensing module as an environmental constraint.

5. The system for real-time monitoring of transpiration of fodder crop according to claim 1, wherein, The refresh frequency of the output value of the transpiration fusion calculation module is consistent with the data update frequency of the multi-source data sensing module.

6. The system for real-time monitoring of transpiration of fodder crop according to claim 1, wherein, The forage crop transpiration real-time monitoring system is also integrated with a data post-processing and quality control unit connected after the transpiration fusion calculation module. The data post-processing and quality control unit analyzes the real-time transpiration monitoring value sequence output by the fusion calculation module in real time and detects and removes outliers based on the sliding time window-based Levey-Jennings criterion.

7. The system according to claim 1, wherein the system is characterized by: The central processing unit of the forage crop transpiration real-time monitoring system is configured with a remote communication interface to upload the final transpiration monitoring results and related stress warning information to a cloud management platform in real time; users can access the cloud platform through authorized terminal devices to view the field transpiration dynamics in real time and receive the irrigation recommendation instructions automatically generated by the system based on the transpiration cumulative value exceeding the preset irrigation threshold.

8. The system according to claim 1, wherein the system is characterized by: All sensors in the multi-source data sensing module gather data through Internet of Things nodes and transmit encrypted data packets to the central processing unit of the system in a wireless communication manner.

9. The system for real-time monitoring of transpiration of fodder crop according to claim 1, wherein, In the stress correction factor calculation program, the crown air temperature difference deviation degree is defined as the absolute value of the difference between the real-time crown air temperature difference and the upper limit of the threshold interval, and the stem flow rate attenuation rate is defined as the absolute value of the ratio of the real-time stem flow rate to the lower limit of the reference interval minus 1.

Images