Dynamic weighing control method in long-strip-shaped bare seedling transplanting process

Abstract

The invention discloses a dynamic weighing control method in a long-strip-shaped bare seedling transplanting process, and relates to the technical field of agricultural intelligent equipment, and the method comprises the steps: S1, obtaining environment humidity data and fog concentration data, judging whether the data exceed a normal range or not through a preset threshold value, obtaining a sensor surface condensation risk assessment result in a high-humidity fog state, and determining whether the data exceed a normal range or not; s2, according to the condensation risk assessment result, activating a sensor surface temperature control mechanism by adopting a heating module, and determining a temperature rise value to prevent formation of water drops and maintain a dry state of the sensor surface; according to the dynamic weighing control method in the long-strip-shaped bare seedling transplanting process, closed-loop control is achieved. The precision and the survival rate of seedling transplanting are remarkably improved, and the agricultural production efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of agricultural intelligent equipment technology, specifically to a dynamic weighing control method for transplanting long, bare seedlings. Background Technology

[0002] Weighing technology is crucial in agricultural production, especially in seedling transplantation, where accurate weight data directly impacts planting decisions and survival rates. Precise measurement of seedling weight allows for optimization of soil composition, irrigation levels, and transplanting timing, thereby improving agricultural efficiency and quality. However, the impact of high humidity and foggy environments on weighing accuracy remains a significant challenge, particularly in greenhouses, coastal areas, or rainy regions. Ensuring the stability and accuracy of weighing equipment in complex environments is key to advancing intelligent and precision agriculture. Existing weighing equipment often performs poorly in high humidity and foggy conditions, primarily due to the susceptibility of sensors to environmental interference. Traditional protective measures, such as simple sealed enclosures, while preventing moisture ingress to some extent, cannot address condensation on sensor surfaces. This condensation not only adds extra weight to the sensor surface but can also alter sensor sensitivity, leading to distorted measurement data. Furthermore, existing equipment typically lacks the ability to dynamically adapt to environmental changes; for example, in the event of rapid humidity fluctuations, protective devices fail to adjust promptly, resulting in decreased accuracy. The core technical challenge lies in effectively preventing the impact of condensation on the sensor while maintaining the device's high sensitivity and stability. First, condensation forms water droplets on the sensor surface, altering its stress state and leading to measurement errors. For example, in greenhouses, when weighing seedlings, the sensor surface may record additional water droplet weight due to condensation, resulting in inflated data and misleading soil mix decisions during transplanting. Second, the sensor needs to maintain sensitivity in high-humidity environments over extended periods, but moisture can penetrate the device, affecting the stability of electronic components and potentially causing inaccurate measurements or even device damage. The contradiction between moisture condensation and sensor sensitivity becomes the key to technological breakthroughs. Summary of the Invention

[0003] The purpose of this invention is to provide a dynamic weighing control method for transplanting long, bare seedlings, thereby solving the problems existing in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a dynamic weighing control method for transplanting elongated bare seedlings, the method comprising the following steps: S1. Acquire ambient humidity data and fog concentration data, and determine whether the data exceeds the normal range by using a preset threshold to obtain the sensor surface condensation risk assessment result under high humidity fog conditions. S2. Based on the condensation risk assessment results, the heating module is used to activate the sensor surface temperature control mechanism to determine the temperature rise value to prevent water droplet formation and maintain the sensor surface dryness. S3. Obtain the initial seedling weight signal from the sensor in the dry state, process the signal through a signal filtering algorithm, determine whether the filtered signal is stable, and obtain the weight data after removing noise interference. S4. For the weight data, a calibration algorithm is used to compare it with a preset standard weight curve, the deviation value is determined and the data output is adjusted to obtain an accurate seedling weight value for transplanting decision-making. S5. Calculate the soil ratio parameters using the precise seedling weight value. If the calculated parameters exceed the optimization threshold, trigger the adjustment command to obtain a ratio correction scheme to match the irrigation optimization requirements. S6. Extract irrigation volume indicators from the revised plan, use a prediction model to simulate the survival rate trend, determine whether the simulated trend reaches the expected level, and determine the final transplanting timing parameters. S7. Integrate all data streams based on the transplanting timing parameters, generate a comprehensive report data packet, and transmit it to the agricultural system interface to complete the closed-loop control of the entire weighing process.

[0005] Preferably, the ambient humidity data and fog concentration data are acquired by an environmental acquisition module located around the sensor. The environmental acquisition module includes a humidity sensor and a fog sensor. The humidity sensor and fog sensor monitor the air humidity and aerosol concentration in real time, respectively, and send the detection results to the central control unit for condensation risk assessment via a wireless communication module.

[0006] Preferably, the heating module includes a flexible heating element and a temperature control unit. The temperature control unit controls the operation of the flexible heating element according to the condensation risk assessment result. The flexible heating element is attached to the outer surface of the weighing sensor to maintain its outer surface temperature above the ambient dew point temperature.

[0007] Preferably, the signal filtering algorithm is a Kalman filter algorithm or a low-pass filter algorithm. The filtering algorithm continuously processes the initial seedling weight signal according to a set time window, removes instantaneous fluctuations, and outputs a relatively stable net weight signal as weight data.

[0008] Preferably, the calibration algorithm includes a regression model based on historical measurement samples. The regression model takes current weight data and environmental parameters as input, outputs the deviation value from the standard weight curve, and calls a preset weight compensation function based on the deviation value to correct the data.

[0009] Preferably, the prediction model is a combination of the support vector regression model (SVR) and the random forest model. The prediction model takes the accurate seedling weight value, the corrected irrigation amount and historical survival rate data as input, and outputs the expected survival rate trend of the seedlings at different transplanting times.

[0010] Preferably, the comprehensive report data package includes environmental parameter data, dynamic weighing data, irrigation ratio parameters, transplanting timing parameters, and predicted survival rate curves. The data package is encapsulated in JSON format and uploaded to the agricultural IoT platform interface via the MQTT protocol.

[0011] Preferably, the temperature rise value is determined by: based on the difference between the current surface temperature of the sensor and the ambient dew point temperature, obtaining the required power supply by looking up the temperature rise-power mapping table, and controlling the heating module to output the corresponding power to achieve temperature regulation.

[0012] Preferably, the standard weight curve is preset according to different seedling types. The seedling types are identified and classified by a visual recognition module. The visual recognition module includes an industrial camera and a seedling image database to automatically match the corresponding weight curve template.

[0013] Preferably, the ratio correction scheme is generated based on the current seedling weight, target growth cycle, seedbed substrate type, and historical ratio optimization data, and the adjustment instruction is generated by the central processing unit based on multi-parameter cross-comparison.

[0014] As can be seen from the above technical solution, the present invention has the following beneficial effects: This dynamic weighing control method for transplanting strip-shaped bare seedlings addresses the complexity of seedling weighing, environmental control, and transplanting decisions in agricultural production, solving the problem of accurate transplanting under the coupled influence of multiple factors such as humidity, fog, weight signal noise, and soil composition. The invention collects environmental humidity and fog concentration data, assesses the risk of condensation on the sensor surface using preset thresholds, activates the heating module to dynamically adjust the temperature, prevents water droplet formation, and ensures the sensor operates dry and stably. Then, it processes the initial seedling weight signal through a signal filtering algorithm to eliminate noise interference, and compares it with a standard weight curve using a calibration algorithm to output accurate weight data. Based on this data, it calculates soil composition parameters, triggers optimization adjustment commands, and generates a composition correction scheme. Finally, it simulates irrigation volume and survival rate trends through a predictive model to determine the optimal transplanting time, integrates the data stream to generate a comprehensive report, and transmits it to the agricultural system interface to achieve closed-loop control. This invention significantly improves the accuracy and survival rate of seedling transplanting, optimizing agricultural production efficiency. Attached Figure Description

[0015] Figure 1 This is a flowchart of the dynamic weighing control method for transplanting long, bare seedlings according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] like Figure 1 As shown, the present invention provides a technical solution: a dynamic weighing control method for transplanting elongated bare seedlings, the method comprising the following steps: S1. Acquire ambient humidity data and fog concentration data, and determine whether the data exceeds the normal range by using a preset threshold to obtain the sensor surface condensation risk assessment result under high humidity fog conditions. S2. Based on the condensation risk assessment results, the heating module is used to activate the sensor surface temperature control mechanism to determine the temperature rise value to prevent water droplet formation and maintain the sensor surface dryness. S3. Obtain the initial seedling weight signal from the sensor in the dry state, process the signal through a signal filtering algorithm, determine whether the filtered signal is stable, and obtain the weight data after removing noise interference. S4. For the weight data, a calibration algorithm is used to compare it with a preset standard weight curve, the deviation value is determined and the data output is adjusted to obtain an accurate seedling weight value for transplanting decision-making. S5. Calculate the soil ratio parameters using the precise seedling weight value. If the calculated parameters exceed the optimization threshold, trigger the adjustment command to obtain a ratio correction scheme to match the irrigation optimization requirements. S6. Extract irrigation volume indicators from the revised plan, use a prediction model to simulate the survival rate trend, determine whether the simulated trend reaches the expected level, and determine the final transplanting timing parameters. S7. Integrate all data streams based on the transplanting timing parameters, generate a comprehensive report data packet, and transmit it to the agricultural system interface to complete the closed-loop control of the entire weighing process.

[0018] This method first acquires environmental humidity and fog concentration data through environmental sensing modules deployed in the transplanting area. Humidity data is obtained via a relative humidity sensor, expressed as a percentage, while fog concentration data is collected in real-time by a scattering fog concentration monitor, expressed as milligrams per cubic meter. The system has preset humidity thresholds of 85% and fog concentration thresholds of 20 milligrams per cubic meter. When either real-time data exceeds this threshold, the system classifies the environment as a high-humidity fog state. At this point, the system automatically activates a condensation risk assessment module. This module compares the current environmental data with the real-time data collected from the sensor materials (thermal conductivity and surface temperature). If the surface temperature is more than 2 degrees Celsius below the ambient dew point temperature, a risk of surface condensation is identified.

[0019] Assuming a condensation risk is present, the system activates a heating module connected to the sensor surface. This module uses a thermistor to control the current, raising the sensor surface temperature through resistance heating. Specifically, the system sets a target surface temperature 5 degrees Celsius higher than the current dew point temperature. The heating module continuously monitors the surface temperature, controlling the heating rate at 0.5 degrees Celsius per second until the target temperature remains constant for at least 5 seconds, confirming the surface is dry. Once dry, the system begins acquiring the seedling weight signal. This weight signal originates from a resistance strain gauge load cell installed below the weighing platform, which outputs a continuous analog voltage signal. The system converts the analog signal to a digital signal at a frequency of 100 times per second using an analog-to-digital converter, continuously acquiring a data sequence of at least 5 seconds.

[0020] Subsequently, the system performs signal filtering on the acquired weight signal. The filtering process uses a moving average filtering algorithm, specifically dividing 500 consecutive data points into 10 groups in chronological order, with each group containing 50 data points. The arithmetic mean of each group is calculated as the representative value for that time period. Then, it is determined whether the difference between two adjacent averages is less than 0.5 grams. If this stability condition is met, the filtering result is considered stable. If not, the next batch of data is collected, and the above steps are repeated until the condition is met.

[0021] Once a stable weight signal is obtained, the system calls the calibration algorithm to compare it with a preset standard weight curve. The standard curve is established based on historical data and corresponds to the target weight range defined by the seedling specifications. For example, for a seedling with a height of 30 cm, the standard weight range is 180 grams to 220 grams. The calibration process compares the current filtered weight with the nearest point on the standard curve. If the deviation is greater than 5 grams, a correction value is calculated. This correction value is obtained by weighting the historical error mean with the current deviation. The system adds this correction value back to the original filtering result to obtain the final accurate seedling weight value.

[0022] The system further calculates the soil mix parameters for transplanting based on the precise seedling weight. The soil mix includes three materials: nutrient soil, peat moss, and perlite, with their proportions determined by the seedling weight. For example, for a seedling weighing 200 grams, the recommended total soil volume is set at 1000 grams, and the initial recommended ratio of the three components is 5:3:2, i.e., 500 grams, 300 grams, and 200 grams respectively. The system compares this recommended ratio with the recommended standard associated with the current actual measured weight. If the calculated value of any component deviates by more than 10%, a mix correction mechanism is triggered. The mix correction is based on the mix adjustment records for the corresponding weight range in the database, combined with the actual deviation ratio, to perform weighted correction and output new component mass values, forming a mix correction scheme.

[0023] The system then extracts the irrigation amount index from the revised plan. This index is the total soil mass multiplied by a set moisture content coefficient. The moisture content coefficient is determined based on the seedling variety and the current ambient temperature. For example, at an ambient temperature of 25 degrees Celsius, the recommended moisture content coefficient for this seedling variety is 0.35. Therefore, if the revised total soil mass is 1050 grams, the corresponding irrigation amount is 367.5 grams. The system inputs this irrigation amount into the survival rate prediction model. This prediction model is a multivariate regression model based on historical samples. The input variables include seedling weight, soil composition, irrigation amount, and ambient temperature. The output variable is the expected survival rate percentage. The system calculates the trend of the survival rate curve under the current parameters. If the predicted survival rate is higher than 90%, the current transplanting plan is considered to have reached the expected level. The current system time, environmental conditions, and parameter combinations are extracted as the final transplanting timing parameters.

[0024] Finally, the system packages all the key data points from the aforementioned processes, including environmental monitoring data, temperature control logs, weight acquisition and correction results, soil mix adjustment schemes, irrigation amounts, and survival rate predictions, into a unified data package. The data package is encapsulated in a binary structured format and uploaded to the central agricultural decision-making platform via the agricultural system communication interface. Upon receiving the data, the platform records, reviews, and provides further instructions, marking the completion of the closed-loop control process for this seedling dynamic weighing control.

[0025] Ambient humidity data and fog concentration data are acquired by an environmental acquisition module located around the sensor. The environmental acquisition module includes a humidity sensor and a fog sensor. The humidity sensor and fog sensor monitor the air humidity and aerosol concentration in real time, respectively, and send the detection results to the central control unit for condensation risk assessment via a wireless communication module.

[0026] In this embodiment, an environmental acquisition module is fixedly installed around the sensor structure of the dynamic weighing control system. The module includes a humidity sensor and a fog sensor, used to collect relative humidity data and aerosol particle concentration in the air, respectively. The humidity sensor is a digital sensor based on the principle of capacitance change, with an accuracy of ±2%, a response time of less than 3 seconds, and a measurement unit of percentage. The fog sensor uses laser scattering sensing technology, with a detection range of 0 to 100 mg / m³ and a response time of less than 2 seconds. Both sensors continuously collect target parameters in the air at a frequency of 5 times per second through a high-speed sampling module, and the detected real-time data is packaged and transmitted through an internally integrated wireless communication module. This wireless communication module uses low-power wireless transmission technology based on the 2.4GHz frequency band, and has automatic channel switching and anti-interference functions to ensure data integrity.

[0027] The collected data is transmitted to the central control unit via a wireless communication module. The central control unit, the core processor of the system, immediately compares the real-time air humidity and aerosol concentration data with preset threshold standards. The humidity threshold is set at 85%, and the aerosol concentration threshold is set at 20 mg / m³. When the data from any sensor exceeds its corresponding threshold, the system determines that the sensor surface is in a high-humidity fog state. This judgment directly serves as the trigger condition for initiating the subsequent condensation risk assessment mechanism and enters the heating module's temperature control process to prevent water droplet condensation from affecting the weighing sensor in a high-humidity environment, thus providing a preliminary guarantee for the sensor's operational stability.

[0028] The heating module includes a flexible heating element and a temperature control unit. The temperature control unit controls the operation of the flexible heating element according to the condensation risk assessment results. The flexible heating element is attached to the outer surface of the weighing sensor to maintain its outer surface temperature above the ambient dew point temperature.

[0029] In this embodiment, a heating module consisting of a flexible heating element and a temperature control unit is provided to address the potential condensation problem on the sensor surface in high humidity environments. The flexible heating element is a resistive heating element made of polyimide or silicone rubber, with a thickness not exceeding 1 mm. It can be tightly fitted to the metal casing surface of the weighing sensor, covering at least 70% of the sensor casing. The temperature control unit includes a thermistor, a temperature control chip, and a control circuit. It collects the temperature data of the heating element surface and compares it with the ambient dew point temperature calculated in real time by the system. The ambient dew point temperature is calculated in real time by the central control unit based on the ambient temperature and relative humidity. When the sensor casing surface temperature is within 3 degrees Celsius below the dew point temperature, the system considers there to be a risk of condensation.

[0030] At this point, the temperature control unit sends a heating command to the flexible heating element, adjusting the current to control the heating element's temperature rise. The heating rate is controlled at 0.3 degrees Celsius per second, and the target temperature is set to be 5 degrees Celsius above the ambient dew point temperature. The heating process continues until the surface temperature reaches the target value and remains stable for more than 10 seconds. Once the system confirms that the sensor housing surface is dry, weighing signal acquisition can begin. During heating, the temperature control unit continuously monitors the surface temperature. If the temperature fluctuation exceeds ±1 degree Celsius or the external humidity changes significantly, the system automatically adjusts the heating current to maintain the set temperature difference, thereby ensuring that the housing surface is always in a non-condensing state, guaranteeing the stability and reliability of the weighing data.

[0031] The signal filtering algorithm is either a Kalman filter algorithm or a low-pass filter algorithm. The filtering algorithm continuously processes the initial seedling weight signal according to a set time window, removes instantaneous fluctuations, and outputs a relatively stable net weight signal as weight data.

[0032] In this embodiment, to improve the accuracy and stability of the weighing data, the system continuously processes the initial seedling weight signal using either a Kalman filter or a low-pass filter algorithm. The initial weight signal is output by the weighing sensor in analog voltage form, and after analog-to-digital conversion, it forms discrete time-series data with a sampling frequency of 100 times per second. The system sets a time window for the filtering algorithm, with a duration of 5 seconds, meaning each batch contains 500 sampling points. The filtering process performs calculations on these 500 data points in chronological order.

[0033] When using the Kalman filter algorithm, the system first sets an initial state estimate and an estimation error covariance. Then, after receiving actual weight observations at each sampling point, it uses a two-step iterative process of prediction and update to correct the current weight state value. The prediction step calculates the prior estimate for the current time step using the state estimate from the previous time step and the control input; the update step corrects the state estimate by weighting the difference between the actual observation and the prior estimate, and finally outputs the current optimal net weight estimate. This process is executed cyclically throughout the entire time window, ultimately outputting a set of smoothed weight curves after filtering, and extracting the mean of the last stable interval as the current net weight data.

[0034] If a low-pass filtering algorithm is used, the system sets the filter cutoff frequency to 1 Hz and applies a first-order discrete low-pass filter to suppress high-frequency interference. This filter effectively smooths out signal abrupt changes and filters out high-frequency fluctuations by weighting the ratio between the current input value and the previous filter output value using weighting coefficients. After processing, continuous data segments with fluctuation amplitudes not exceeding 0.5 grams in the last few seconds are extracted, their average value is calculated, and the output is the final net weight signal.

[0035] The entire filtering process includes a stability assessment module. The assessment condition is that the fluctuation amplitude of the output signal is less than 0.5 grams within 3 consecutive seconds. If the condition is met, the system considers the net weight data to be stable and reliable, and it can be used for subsequent weight calibration and decision calculations. If the condition is not met, the system continues to collect data from the next time window and repeats the processing until the stability standard is met.

[0036] The calibration algorithm includes a regression model based on historical measurement samples. The regression model takes current weight data and environmental parameters as input, outputs the deviation value from the standard weight curve, and calls a preset weight compensation function to correct the data based on the deviation value.

[0037] In this embodiment, after completing the initial weight signal acquisition and filtering, the dynamic weighing system further calls a calibration algorithm to finely correct the acquired net weight data, thereby improving weighing accuracy and eliminating systematic errors caused by environmental changes. The calibration algorithm is built upon a large number of historical measurement samples and employs a regression modeling method to statistically model relevant parameters included in the samples, such as the actual weight of the seedlings, the weighing output value, ambient temperature, humidity, and wind speed. This model uses the current filtered net weight data and real-time environmental parameters as input variables, and estimates the deviation between the current data point and the standard weight curve using linear or multinomial regression methods.

[0038] The standard weight curve is pre-set based on the seedling variety and specifications, defining the expected weight range for a specific seedling height or diameter. For example, for a bare seedling with a height of 30 cm, the standard weight curve varies between 190 grams and 210 grams. The deviation value output by the regression model represents the difference between the current weighing data and this standard range, expressed in grams. The system calls a preset weight compensation function based on this deviation value. This function is a piecewise linear function structure, setting the correction range according to different deviation ranges. For example, deviations within ±5 grams are treated as the original value; deviations between ±5 grams and ±15 grams are corrected at 80% of the actual deviation; and deviations greater than ±15 grams are limited to a maximum correction of 12 grams to prevent over-correction from amplifying the error.

[0039] The data correction process is executed automatically by the system. Specifically, the current net weight data is added to the output value of the correction function to generate the final weight value. After correction, the system will also perform a boundary comparison with the standard curve. If the corrected data is still outside the standard range, the system records the data as an abnormal weight and issues a prompt signal for subsequent manual verification or automatic retesting. The entire correction process takes no more than 1 second and is suitable for real-time dynamic weighing processes.

[0040] The prediction model is a combination of the support vector regression model (SVR) and the random forest model. The prediction model takes the accurate seedling weight value, the corrected irrigation amount and historical survival rate data as input, and outputs the expected survival rate trend of seedlings under different transplanting times.

[0041] In this embodiment, to improve the scientific rigor and timeliness of seedling transplanting decisions, after acquiring weighing data and correcting soil mix ratios, the system introduces an ensemble prediction model to assess the expected survival rate trends of seedlings under different transplanting time conditions. This prediction model consists of two parts: a support vector regression model and a random forest model. The support vector regression model is suitable for regression prediction with small samples and high-dimensional features, and can capture the nonlinear relationship between seedling physiological indicators and survival rate; the random forest model has good anti-overfitting ability and feature importance recognition ability, and can handle scenarios with large fluctuations in environmental variables. The system employs a weighted ensemble strategy to fuse the outputs of the two models, improving the overall prediction stability and accuracy.

[0042] The model input parameters include accurate seedling weight, corrected irrigation amount, and historical survival rate data. The accurate seedling weight is obtained from the final weight result after filtering and calibration, in grams; the corrected irrigation amount is the amount of irrigation water calculated by the system based on the corrected ratio, in grams; and the historical survival rate data is obtained from statistics on the survival of seedlings of the same variety and specification at different transplanting times in the database, in percentages. These three input parameters are standardized and then input into the SVR and Random Forest models respectively. The SVR model outputs a continuous survival rate prediction curve under current conditions, while the Random Forest model outputs distributed survival rate estimates for multiple possible time points. The system fuses the outputs of the two models according to empirical weights, with the SVR model weight set to 0.6 and the Random Forest model weight set to 0.4.

[0043] The fusion output is a trend curve of expected survival rate, with the horizontal axis representing candidate transplanting times within the next few days and the vertical axis representing the predicted survival rate at the corresponding time points. The system further analyzes the trend curve to identify the time point corresponding to the maximum survival rate, returning it as the parameter for the current optimal transplanting time, and using it to generate subsequent transplanting suggestion reports or automatic control commands. This prediction process is completed by the central control unit, with an average processing time of no more than 2 seconds, enabling rapid response and prediction support for dynamic field data.

[0044] The comprehensive report data package includes environmental parameter data, dynamic weighing data, irrigation ratio parameters, transplanting timing parameters, and predicted survival rate curves. The data package is encapsulated in JSON format and uploaded to the agricultural IoT platform interface via the MQTT protocol.

[0045] In this embodiment, to achieve closed-loop data management of the dynamic weighing process of seedlings, the system integrates all key data into a unified data package after completing data collection, processing, and decision calculation at each stage. This data package contains five core components: the first is environmental parameter data, including real-time collected air temperature, relative humidity, fog concentration, and calculated dew point temperature; the second is dynamic weighing data, including the original weight signal, filtered net weight, calibration correction value, and final confirmed weight value; the third is irrigation ratio parameters, specifically recording the mass ratio of nutrient soil, peat moss, and perlite, and the corresponding calculated irrigation water volume; the fourth is transplanting timing parameters, including the optimal transplanting time determined by the system and its corresponding timestamp; and the fifth is a predicted survival rate curve, listing the predicted survival rates for multiple consecutive candidate transplanting days in a time series, forming complete trend data.

[0046] All the above data is packaged by the central control unit after each weighing control process, and the encapsulation format is a standard JSON structure. Each field adopts a clear key-value pair format, and the field name follows the naming convention of lowercase letters and underscores, such as "env_temperature", "net_weight", "irrigation_ratio", "transplant_time", "survival_rate_curve", etc., to ensure data structure consistency and platform parsing compatibility.

[0047] The encapsulated JSON data packet is uploaded by the communication module integrated into the system. The communication process uses the MQTT protocol, a lightweight message publish-subscribe mechanism characterized by low bandwidth consumption, high reliability, and QoS level support, making it suitable for the communication needs of numerous low-power devices in agricultural scenarios. The data packet is published to the configured MQTT topic path, corresponding to the data acquisition interface node of the agricultural IoT platform. Upon receiving the data, the platform automatically parses the JSON structure and writes it to the database, simultaneously triggering subsequent tasks such as data visualization, expert system suggestion generation, or automated execution commands, enabling remote collaborative management of the entire system.

[0048] The temperature rise value is determined by: based on the difference between the current surface temperature of the sensor and the ambient dew point temperature, obtaining the required power supply by looking up the temperature rise-power mapping table, and controlling the heating module to output the corresponding power to achieve temperature regulation.

[0049] In this embodiment, to achieve precise control of the sensor surface temperature, the system calculates the required temperature rise for the sensor surface after determining there is a risk of condensation. The specific process is as follows: The central control unit first collects the current surface temperature of the sensor housing in real time using a temperature sensor, and then calculates the dew point temperature of the current environment using a humidity conversion formula, combined with the ambient temperature and relative humidity. Subsequently, the system calculates the difference between the dew point temperature and the sensor surface temperature. If the difference is less than 5 degrees Celsius, the system determines that a temperature increase is needed, and the temperature rise is set as the sum of this difference and a safety margin, which is fixed at 3 degrees Celsius.

[0050] Once the temperature rise is obtained, the system consults a pre-established temperature rise-power mapping table, which is generated based on experimental data of the heating module structure, thermal conductivity material properties, and heat transfer efficiency under the operating environment. The mapping table lists the required power supply under different ambient temperatures and target temperature rise values. For example, when the ambient temperature is 20 degrees Celsius and the target temperature rise is 8 degrees Celsius, the mapping table indicates a required power of 4 watts. Based on the found power value, the system controls the PWM modulation signal to adjust the actual output power of the flexible heating element. The heating module employs a constant power control mode, continuously operating at the target power level until the sensor surface temperature probe reports that the surface temperature has reached the target value and remained stable for more than 5 seconds.

[0051] During this process, the temperature control unit collects surface temperature data once per second and judges the matching relationship between the temperature rise rate and the power response in real time. If the actual temperature rise rate deviates from the theoretical value by more than 20%, the output power is automatically adjusted, and this adjustment behavior is recorded for subsequent mapping table optimization. The entire temperature control process ensures that the heating module can quickly raise the sensor surface temperature to a safe level while prioritizing efficiency and controlling energy consumption, thereby effectively suppressing the risk of condensation.

[0052] Standard weight curves are preset according to different seedling types. The seedling types are identified and classified by a visual recognition module, which includes an industrial camera and a seedling image database to automatically match the corresponding weight curve template.

[0053] In this embodiment, to ensure that the standard weight curve used in the calibration algorithm accurately matches the seedling type, a visual recognition module is introduced to automatically classify the seedlings. This module consists of an image acquisition system composed of an industrial camera installed above the weighing channel. The industrial camera is selected with an image resolution of no less than 5 million pixels and a frame rate of more than 30 frames per second, and has automatic exposure and light compensation functions to adapt to changes in field lighting. Before the seedling is placed on the weighing platform, the system initiates the visual recognition process, and the industrial camera takes one image each of the front and side of the seedling. The images are then uploaded to the central control unit for analysis.

[0054] The central control unit accesses a locally deployed seedling image database, which contains feature image datasets and corresponding standard weight curve templates for common seedling species (such as red maple, camphor tree, and ginkgo). The system extracts features from the acquired images, including seedling outline shape, leaf texture, branch structure, and other parameters, and performs classification based on a convolutional neural network model. The model outputs the seedling species label with the highest recognition probability, and the system matches it to a preset standard weight curve template. The standard weight curve uses different specifications (such as seedling height and diameter) of the seedling species as independent variables, outputting corresponding weight ranges as calibration references.

[0055] After completing the identification and curve matching, the system binds the weight template to the sample in the current weighing process and uses it as the baseline for subsequent calibration and deviation calculation. This ensures that each seedling is compared with the standard curve corresponding to its actual species, avoiding weighing errors or calibration failures caused by incorrect seedling species identification. The entire image recognition and template matching process takes an average of no more than 2 seconds, meeting the real-time requirements of dynamic transplanting processes.

[0056] The generation of the ratio correction scheme is based on the current seedling weight, target growth cycle, seedbed substrate type and historical ratio optimization data. The adjustment instructions are generated by the central processing unit based on multi-parameter cross-comparison.

[0057] In this embodiment, to achieve precise soil ratio management during seedling transplantation, the system, based on the initial calculation of standard ratio parameters, further introduces a ratio correction mechanism to optimize and adjust the original ratio scheme. This correction process is executed by the central processing unit, and its core basis includes four parameters: current seedling weight, target growth cycle, seedbed substrate type, and historical ratio optimization data. The current seedling weight is output by the dynamic weighing module and is the final accurate seedling weight after calibration; the target growth cycle is input by the agricultural management system or user, in days or weeks, and is used to define the length of time required for seedlings to grow in the seedbed; the seedbed substrate type is input through settings, including common peat moss, vermiculite, coconut coir, perlite, etc., and different substrates have differences in water retention, aeration, and nutrient slow-release capacity; historical ratio optimization data comes from the database recording the soil ratio and actual growth effect evaluation records of similar seedlings under similar environmental and periodic conditions.

[0058] After receiving the above parameters, the central processing unit first performs a combined analysis of the current weight value and the target cycle to determine the seedlings' demand for water and nutrients, and compares the physical properties of the corresponding substrate to preliminarily assess its suitability. Based on this, the system selects successful samples matching the conditions from historical mix optimization data, extracts the corresponding soil ternary ratio schemes and their key indicators such as survival rate and root development score, and performs weighted summaries to generate a recommended mix range. The system then cross-compares the current initial mix scheme with this recommended range. If any component (such as peat moss or perlite) exceeds the recommended range, a mix correction suggestion is generated, including the name of the component to be adjusted, the adjustment range (in grams), and the direction of adjustment (increase or decrease). All comparisons and suggestion generation processes are completed based on a rule engine and statistical optimization algorithms to ensure that the suggested schemes are feasible and have expected optimization.

[0059] Ultimately, the system generates a complete proportioning correction scheme and converts it into a standardized adjustment instruction. The instruction includes fields such as target component, adjustment value, execution sequence number, and instruction check code. This instruction is synchronously transmitted to the filling control module or manual prompting system to achieve automatic or assisted execution of the correction operation. The entire comparison and correction process has a response time of no more than 1 second, enabling real-time proportioning adjustments during dynamic weighing.

[0060] 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 dynamic weighing control method for transplanting long, bare seedlings, characterized in that, The method includes the following steps: S1. Acquire ambient humidity data and fog concentration data, and determine whether the data exceeds the normal range by using a preset threshold to obtain the sensor surface condensation risk assessment result under high humidity fog conditions. S2. Based on the condensation risk assessment results, the heating module is used to activate the sensor surface temperature control mechanism to determine the temperature rise value to prevent water droplet formation and maintain the sensor surface dryness. S3. Obtain the initial seedling weight signal from the sensor in the dry state, process the signal through a signal filtering algorithm, determine whether the filtered signal is stable, and obtain the weight data after removing noise interference. S4. For the weight data, a calibration algorithm is used to compare it with a preset standard weight curve, the deviation value is determined and the data output is adjusted to obtain an accurate seedling weight value for transplanting decision-making. S5. Calculate the soil ratio parameters using the precise seedling weight value. If the calculated parameters exceed the optimization threshold, trigger the adjustment command to obtain a ratio correction scheme to match the irrigation optimization requirements. S6. Extract irrigation volume indicators from the revised plan, use a prediction model to simulate the survival rate trend, determine whether the simulated trend reaches the expected level, and determine the final transplanting timing parameters. S7. Integrate all data streams based on the transplanting timing parameters, generate a comprehensive report data packet, and transmit it to the agricultural system interface to complete the closed-loop control of the entire weighing process.

2. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The ambient humidity data and fog concentration data are acquired by an environmental acquisition module located around the sensor. The environmental acquisition module includes a humidity sensor and a fog sensor. The humidity sensor and fog sensor monitor the air humidity and aerosol concentration in real time, respectively, and send the detection results to the central control unit for condensation risk assessment via a wireless communication module.

3. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The heating module includes a flexible heating element and a temperature control unit. The temperature control unit controls the operation of the flexible heating element according to the condensation risk assessment result. The flexible heating element is attached to the outer surface of the weighing sensor to maintain its outer surface temperature above the ambient dew point temperature.

4. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The signal filtering algorithm is either a Kalman filter algorithm or a low-pass filter algorithm. The filtering algorithm continuously processes the initial seedling weight signal according to a set time window, removes instantaneous fluctuations, and outputs a relatively stable net weight signal as weight data.

5. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The calibration algorithm includes a regression model based on historical measurement samples. The regression model takes current weight data and environmental parameters as input, outputs the deviation value from the standard weight curve, and calls a preset weight compensation function to correct the data based on the deviation value.

6. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The prediction model is a combination of the support vector regression model (SVR) and the random forest model. The prediction model takes the accurate seedling weight value, the corrected irrigation amount and the historical survival rate data as input, and outputs the expected survival rate trend of seedlings under different transplanting times.

7. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The comprehensive report data package includes environmental parameter data, dynamic weighing data, irrigation ratio parameters, transplanting timing parameters, and predicted survival rate curves. The data package is encapsulated in JSON format and uploaded to the agricultural IoT platform interface via the MQTT protocol.

8. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The temperature rise value is determined by: based on the difference between the current surface temperature of the sensor and the ambient dew point temperature, obtaining the required power supply by looking up the temperature rise-power mapping table, and controlling the heating module to output the corresponding power to achieve temperature regulation.

9. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The standard weight curves are preset according to different seedling types. The seedling types are identified and classified by a visual recognition module, which includes an industrial camera and a seedling image database to automatically match the corresponding weight curve template.

10. The dynamic weighing control method for transplanting elongated bare seedlings according to claim 1, characterized in that: The formula adjustment scheme is generated based on the current seedling weight, target growth cycle, seedbed substrate type and historical formula optimization data. The adjustment command is generated by the central processing unit based on multi-parameter cross-comparison.

Images