An adaptive fixed-point quantitative fertilization control system and control method

Abstract

This invention belongs to the field of fertilization control technology, specifically relating to an adaptive fixed-point quantitative fertilization control system and method, including the following steps: S1, acquiring the real-time position, speed, and heading angle of the agricultural machinery, and simultaneously collecting multispectral images of the crop canopy and soil electrical conductivity data at the current position; extracting the crop canopy spectral index from the crop canopy multispectral image, and constructing a spatiotemporal influence weight matrix representing the spatial correlation of crop nutrient requirements based on the historical and current crop canopy spectral indices of multiple sampling points within the current position and a preset spatial neighborhood; S2, calculating a position correction vector to compensate for the spatial lag between the sensor measurement position and the fertilization actuator's action position, based on the agricultural machinery's speed and heading angle. This invention achieves dual precise correction of application position and application amount, ensuring that fertilizer can be accurately applied to the target point, and improving the positional and dosage accuracy of fixed-point quantitative fertilization operations.

Description

Technical Field

[0001] This invention belongs to the field of fertilization control technology, specifically relating to an adaptive fixed-point and quantitative fertilization control system and control method. Background Technology

[0002] Precision agriculture is the core direction of modern agricultural development, and one of its key technologies is variable-rate fertilization, which involves applying fertilizer in real time and in variable quantities based on the actual growth status and nutrient requirements of crops in different locations in the field. Traditional variable-rate fertilization technology assesses crop nitrogen nutrition status by calculating spectral indices such as the normalized difference in vegetation index (NDPI) of the crop canopy, and combines this with information on soil texture and moisture reflected by soil electrical conductivity to jointly determine the amount of fertilizer to be applied.

[0003] Most existing technologies treat each sampling point as an independent decision-making unit, ignoring the spatial continuity and correlation of crop growth, as well as the characteristics of nutrient requirements changing over time. This approach may lead to biased fertilization decisions due to noise or anomalies in single-point data, failing to fully utilize the inherent spatial heterogeneity within the field. Furthermore, on high-speed agricultural machinery platforms, there is a delay between sensor perception and fertilization actuator response, mainly including the lag between the sensor measurement position and the actuator's action position, as well as the mechanical response delay of the actuator itself. Traditional control methods often lack sufficient precision in compensating for these complex delays and are ill-suited to varying machinery speeds and other operating conditions. Therefore, there is an urgent need for a fixed-point, quantitative fertilization control method that comprehensively considers the spatiotemporal correlation of crop nutrient requirements, accurately compensates for delays in multi-source systems, and makes decisions based on a nonlinear model that better aligns with agronomic principles. This would further improve fertilization accuracy, increase fertilizer utilization, and reduce agricultural non-point source pollution. Summary of the Invention

[0004] This invention provides an adaptive fixed-point and quantitative fertilization control system and method to solve the technical problems of existing variable fertilization that ignore the spatiotemporal correlation of crop nutrients and have insufficient system delay compensation, requiring precise decision-making methods.

[0005] In a first aspect, the present invention provides an adaptive method for controlling fixed-point and quantitative fertilization, comprising the following steps:

[0006] S1: Obtain the real-time position, speed, and heading angle of the agricultural machinery, and simultaneously collect multispectral images of the crop canopy and soil electrical conductivity data at the current position; extract the crop canopy spectral index from the crop canopy multispectral image, and construct a spatiotemporal influence weight matrix representing the spatial correlation of crop nutrient demand based on the historical and current crop canopy spectral indices of multiple sampling points within the current position and preset spatial neighborhood;

[0007] S2, based on the speed and heading angle of the agricultural machinery, calculate the position correction vector to compensate for the spatial lag between the sensor measurement position and the fertilizer actuator action position; take the crop canopy spectral index and soil electrical conductivity data at the current position as input features, use the spatiotemporal influence weight matrix for weighted fusion to obtain the comprehensive state parameter, and substitute the comprehensive state parameter into the preset double S-shaped saturated response function to calculate the basic fertilizer requirement at the current position;

[0008] S3, based on the basic fertilization demand, corrects the preset fertilization actuator flow-response delay model, and combines it with the phase advance compensation algorithm to convert the basic fertilization demand into a fertilization control signal.

[0009] S4 combines the fertilization control signal and the position correction vector to generate a fertilization execution command, which drives the fertilization actuator to perform fertilization operations at the calibrated target position.

[0010] Furthermore, a spatiotemporal influence weight matrix representing the spatial correlation of crop nutrient requirements is constructed, including:

[0011] Centered on the current location, set a spatial neighborhood grid of a preset size; obtain all sampling points within the neighborhood at the current time. and historical moments The crop canopy spectral index, j=1, 2, ..., T;

[0012] The spatial weights of each neighboring point i and the center point are calculated using the Gaussian decay function. ,in Let i be the Euclidean distance between the neighboring point i and the center point. For spatial attenuation scale parameters;

[0013] Historical moments are calculated using an exponential decay function. Relative to the current time Time weight ,in This is the time decay coefficient;

[0014] Multiplying the spatial weight by the temporal weight yields the spatiotemporal influence weight of each neighborhood sampling point at different times. They are combined into a spatiotemporal influence weight matrix.

[0015] Furthermore, based on the speed and heading angle of the agricultural machinery, a position correction vector is calculated to compensate for the spatial lag between the sensor measurement position and the fertilizer actuator's operating position, including:

[0016] Obtain the fixed physical distance d between the sensor and the fertilizer actuator, as well as the inherent response delay time. The real-time speed v of the agricultural machinery is compared with the response delay time. Multiplication yields the delayed displacement Add the physical distance d to the delayed displacement s to obtain the total distance D = d + s; combine this with the real-time heading angle. Calculate the correction components of the correction vector in the east-west direction. and the correction components in the north-south direction , forming a position correction vector .

[0017] Furthermore, weighted fusion is performed using the spatiotemporal influence weight matrix to obtain the comprehensive state parameters, including:

[0018] All crop canopy spectral indices within the spatial neighborhood and from historical periods, along with soil electrical conductivity data at the current location, were subjected to min-max normalization to the [0, 1] interval. The spatiotemporal weighted spectral indices were then weighted and summed using a spatiotemporal influence weight matrix to obtain the spatiotemporal weighted spectral index. Spatiotemporal weighted spectral index Compared with normalized soil electrical conductivity data According to the preset weighting coefficients and Perform linear summation to obtain the comprehensive state parameters. ,in .

[0019] Furthermore, by substituting the comprehensive state parameters into the preset double S-shaped saturated response function, the basic fertilizer requirement at the current location is calculated, including:

[0020] Set the crop nutrient stress threshold for the comprehensive state parameter S. With nutrient adequacy threshold ;when At that time, the basic fertilizer requirement is the preset maximum fertilizer application rate. ;when At that time, the basic fertilizer requirement is the preset minimum fertilizer amount. ;when At that time, the basic fertilizer requirement F is calculated using the following formula:

[0021]

[0022] Where k is the curve steepness coefficient. In response to the midpoint value, and .

[0023] Furthermore, based on the basic fertilization demand, the preset fertilization actuator flow-response delay model is corrected, and combined with a phase lead compensation algorithm, the basic fertilization demand is converted into a fertilization control signal, including:

[0024] Based on the basic fertilization requirement, the preset flow-response delay model of the fertilization actuator is corrected; and based on the corrected flow-response delay model, a transfer function is adopted. (in The phase lead compensation algorithm of ) is as follows, where, This is the gain coefficient. and It is a time constant. The Laplace operator converts the basic fertilizer requirement into a compensated control quantity; the compensated control quantity is then converted into a duty cycle setpoint to generate a PWM signal, which serves as the fertilizer control signal.

[0025] Furthermore, by combining the fertilization control signal and the position correction vector, fertilization execution instructions are generated, including:

[0026] Current real-time GPS location coordinates With position correction vector Perform vector addition to obtain the coordinates of the target fertilization location. ;

[0027] The target fertilization location coordinates and the PWM duty cycle value determined by the fertilization control signal are packaged into a data frame and sent as a fertilization execution command to the fertilization actuator controller.

[0028] Furthermore, in S1, the real-time position is obtained through RTK-GNSS, the heading angle is obtained through IMU, and the velocity is calculated from the rate of change of the GNSS coordinate points over time.

[0029] Furthermore, in S3, the fertilization control signal is an analog voltage of 0 to 5 volts.

[0030] Secondly, the present invention provides an adaptive fixed-point quantitative fertilization control system, including a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned adaptive fixed-point quantitative fertilization control method is implemented.

[0031] The beneficial effects are as follows: By constructing a spatiotemporal influence weight matrix, this invention comprehensively considers the spatial correlation and temporal continuity of crop nutrient requirements, overcoming the biases and instabilities that may arise from relying solely on single-point information for decision-making. This results in smoother and more reliable nutrient requirement assessments, better aligning with the actual continuous distribution patterns of crop growth in the field. The invention employs a double S-shaped saturated response function to calculate fertilizer application rates. This nonlinear model more closely reflects the agronomic principles of crop nutrient absorption, avoiding over- or under-fertilization caused by linear models, making fertilization decisions more scientific and improving fertilizer utilization. Furthermore, this invention utilizes a control strategy combining position correction and phase lead compensation to address both the spatial lag between the sensor and actuator and the response delay of the fertilization actuator itself. This achieves precise dual correction of application location and application rate, ensuring accurate application of fertilizer to the target point and improving the positional and dosage accuracy of fixed-point, quantitative fertilization operations. Attached Figure Description

[0032] Figure 1 This is a flowchart of an adaptive fixed-point and fixed-quantity fertilization control method. Detailed Implementation

[0033] 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, not all, of the embodiments of the present invention. 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.

[0034] An embodiment of the adaptive fixed-point and quantitative fertilization control method provided by the present invention:

[0035] like Figure 1 As shown, the adaptive fixed-point and fixed-quantity fertilization control method includes the following steps:

[0036] S1: Obtain the real-time position, speed, and heading angle of the agricultural machinery, and simultaneously collect multispectral images of the crop canopy and soil electrical conductivity data at the current position; extract the crop canopy spectral index from the crop canopy multispectral image, and construct a spatiotemporal influence weight matrix representing the spatial correlation of crop nutrient demand based on the historical and current crop canopy spectral indices of multiple sampling points within the current position and preset spatial neighborhood.

[0037] Specifically, latitude and longitude coordinates are acquired by a real-time differential global navigation satellite system (RTK-GNSS) receiver installed on the agricultural machinery, and heading angle information is obtained by combining this with an inertial measurement unit (IMU). Velocity information is calculated from the rate of change of GNSS coordinates over time. A multispectral camera installed in front of the agricultural machinery captures images containing near-infrared, red-edge, red, green, and blue bands at fixed time intervals or distance intervals. A contact or non-contact soil conductivity sensor is towed or installed under the agricultural machinery and is hardware-synchronized with the multispectral camera via a unified pulse-per-second signal to ensure that each frame of image and each conductivity reading precisely corresponds to a geographic coordinate.

[0038] Radiometric correction and image stitching were performed on the acquired multispectral images, and then the Normalized Difference Vegetation Index (NDVI) was calculated for each pixel. To represent the crop status at the current location, the average NDVI value of the corresponding image region was extracted. A spatial neighborhood of, for example, 5m × 5m was defined, and all NDVI sampling points within this neighborhood, collected in the past two operations and in the current operation, were retrieved from the historical database. Based on the Gaussian kernel function, weights were calculated according to the time interval and spatial distance between each neighborhood sampling point and the current point; points with longer time intervals and greater spatial distances had lower weights. The set of weights for all neighborhood points constituted the spatiotemporal influence weight matrix.

[0039] In an optional embodiment, a spatiotemporal influence weight matrix representing the spatial correlation of crop nutrient requirements is constructed, including:

[0040] Centered on the current location, set a spatial neighborhood grid of a preset size; obtain all sampling points within the neighborhood at the current time. and historical moments The crop canopy spectral index, j=1, 2, ..., T;

[0041] The spatial weights of each neighboring point i and the center point are calculated using the Gaussian decay function. ,in Let i be the Euclidean distance between the neighboring point i and the center point. For spatial attenuation scale parameters;

[0042] Historical moments are calculated using an exponential decay function. Relative to the current time Time weight ,in This is the time decay coefficient;

[0043] Multiplying the spatial weight by the temporal weight yields the spatiotemporal influence weight of each neighborhood sampling point at different times. They are combined into a spatiotemporal influence weight matrix.

[0044] Specifically, for example, a 10m × 10m square grid is defined as the neighborhood, centered on the current location of the agricultural machinery. This grid includes the center point and eight other historical data sampling points. Normalized Difference Vegetation Index (NDI) data for these nine points are retrieved at the current time and two previous times, such as 5 minutes and 10 minutes ago. A total of 27 data points are obtained, each corresponding to a specific spatial location and time point.

[0045] For spatial weights, assuming a sampling point in the neighborhood is 3m from the center point and the spatial decay scale parameter is set to 5m, the spatial weight is calculated using a Gaussian function and is approximately 0.835, indicating that points closer to the center point have a greater influence. For temporal weights, assuming a time decay coefficient of 0.1, the time weight of data from 5 minutes ago is approximately 0.607, while the weight of data from 10 minutes ago drops to approximately 0.368, reflecting the principle that data from further back has a lower influence. Multiplying the spatial weight of each point by the corresponding temporal weight yields the spatiotemporal influence weight of that data point. For example, the spatiotemporal weight of the point 3m away 5 minutes ago is approximately 0.507. Calculating the spatiotemporal weights of all 27 data points constitutes a complete spatiotemporal influence weight matrix.

[0046] S2, based on the speed and heading angle of the agricultural machinery, calculate the position correction vector to compensate for the spatial lag between the sensor measurement position and the fertilizer actuator action position; take the crop canopy spectral index and soil electrical conductivity data at the current position as input features, use the spatiotemporal influence weight matrix for weighted fusion to obtain the comprehensive state parameters, and substitute the comprehensive state parameters into the preset double S-shaped saturated response function to calculate the basic fertilizer requirement at the current position.

[0047] Specifically, the fixed physical offset between the center point of the multispectral camera's field of view and the fertilizer actuator nozzle or furrow opener in the agricultural machinery coordinate system is measured and calibrated in advance. This physical offset includes the longitudinal distance. and lateral distance Based on the real-time heading angle of the agricultural machinery By performing coordinate rotation transformation, this fixed physical offset is converted into a position correction vector in the geodetic coordinate system. For example, the eastward correction component is... The northward correction component is .

[0048] Multiply and sum all NDVI values ​​at the current location and within its neighborhood by the spatiotemporal influence weight matrix to obtain a spatiotemporally smoothed NDVI value. Then, multiply this smoothed NDVI value by the soil conductivity at the current location. Through a linear or nonlinear fusion model, such as a comprehensive state parameter Where a and b are empirical coefficients, a comprehensive index that fully reflects crop growth and soil environment is obtained. This comprehensive state parameter S is then substituted into a form resembling fertilizer application rate. of The bisigma function of the power of 1, where and For the maximum and minimum fertilizer application rates, k and The shape parameters of the response curve are used to calculate the basic fertilizer requirement that conforms to the agronomic saturation response law.

[0049] To calculate the precise geographical location at which the fertilization command needs to be applied in advance, in one optional embodiment, a position correction vector is calculated based on the speed and heading angle of the agricultural machinery to compensate for the spatial lag between the sensor measurement location and the fertilization actuator's application location, including:

[0050] Obtain the fixed physical distance d between the sensor and the fertilizer actuator, as well as the inherent response delay time. The real-time speed v of the agricultural machinery is compared with the response delay time. Multiplication yields the delayed displacement Add the physical distance d to the delayed displacement s to obtain the total distance D = d + s; combine this with the real-time heading angle. Calculate the correction components of the correction vector in the east-west direction. and the correction components in the north-south direction , forming a position correction vector .

[0051] Specifically, two fixed parameters are determined: the sensor is installed 2.5m in front of the fertilizer actuator, which is the physical distance d; and the response delay time from when the system issues a command to when the actuator actually sprays fertilizer. The time is 0.8s. Obtain the current driving status of the agricultural machinery, such as real-time speed v = 2m / s and heading angle. It is 60°, with due north at 0°.

[0052] Due to the system response delay, the farm machinery will continue to travel a certain distance within 0.8 seconds. This delayed displacement s is equal to the speed multiplied by the delay time, which is 1.6m. The total distance D is the sum of the physical distance d and the delayed displacement s, which is 4.1m. The data measured by the sensor at the current location is used to determine the fertilizer application amount at points 4.1m away. The total distance is decomposed onto geographic coordinates. Based on a heading angle of 60°, the east-west correction component ΔE is calculated to be approximately 3.55m; the north-south correction component ΔN is calculated to be 2.05m.

[0053] To integrate multi-source information for a comprehensive assessment of the nutrient status of a land parcel, in an optional embodiment, a weighted fusion is performed using a spatiotemporal influence weight matrix to obtain a comprehensive state parameter, including:

[0054] All crop canopy spectral indices within the spatial neighborhood and from historical periods, along with soil electrical conductivity data at the current location, were subjected to min-max normalization to the [0, 1] interval. The spatiotemporal weighted spectral indices were then weighted and summed using a spatiotemporal influence weight matrix to obtain the spatiotemporal weighted spectral index. Spatiotemporal weighted spectral index Compared with normalized soil electrical conductivity data According to the preset weighting coefficients and Perform linear summation to obtain the comprehensive state parameters. ,in .

[0055] Specifically, all input data undergoes standardization to eliminate dimensional differences. For example, assuming the NDVI values ​​collected in the field range from 0.2 to 0.9, a measured NDVI value of 0.6 would have a normalized value of 0.571. Similarly, if the soil electrical conductivity data (EC) values ​​range from 0.5 to 2.5 dS / m, a measured EC value of 1.5 dS / m would have a normalized value of 0.571. It's 0.5. The normalization process maps all spectral and soil data into a uniform range of 0 to 1.

[0056] Using the previously calculated spatiotemporal influence weight matrix, a weighted sum is performed on all normalized spectral indices. The normalized value of each spectral index is multiplied by its corresponding spatiotemporal weight, and then all products are summed to obtain a comprehensive spatiotemporal weighted spectral index. Assuming the calculated result is 0.65, the value representing crop growth will be... and representing soil characteristics A linear combination is performed. With a weighting coefficient α of 0.7 for crop growth and a weighting coefficient β of 0.3 for soil properties, the calculated comprehensive state parameter S is 0.605. The S value represents a comprehensive evaluation of the nutrient requirements at the current location.

[0057] In an optional embodiment, the comprehensive state parameters are substituted into a preset double S-shaped saturated response function to calculate the basic fertilizer requirement at the current location, including:

[0058] Set the crop nutrient stress threshold for the comprehensive state parameter S. With nutrient adequacy threshold ;when At that time, the basic fertilizer requirement is the preset maximum fertilizer application rate. ;when At that time, the basic fertilizer requirement is the preset minimum fertilizer amount. ;when At that time, the basic fertilizer requirement F is calculated using the following formula:

[0059]

[0060] Where k is the curve steepness coefficient. In response to the midpoint value, and .

[0061] Specifically, this method is used to convert abstract, comprehensive state parameters into specific fertilizer application rates. Key parameters are set based on agronomic knowledge. For example, a maximum fertilizer application rate is set. Minimum fertilizer application rate: 200 kg / ha The target is 30 kg / ha. A threshold for the comprehensive state parameter S is also defined; when S falls below 0.3, it is considered severe nutrient stress. When S is above 0.8, it is considered that nutrients are sufficient. The curve steepness coefficient k is set to 12, and the response midpoint value is... The value is 0.55.

[0062] Substitute the comprehensive state parameter S calculated in the previous step, for example, 0.605, into the decision logic. Because 0.605 is within... Values ​​0.3 and Since the value is between 0.8 and 0.8, a double S-shaped function is needed for calculation. Substituting all parameters into the formula: the basic fertilizer requirement F = 30 + 170 / 1, and the sum of an exponential term whose exponent is the difference between 12 × 0.605 and 0.55. The calculated result is approximately 87.8 kg / ha. If the calculated S value is 0.2, it is lower than... If the S value is 0.9, then the maximum fertilizer application rate of 200 kg / ha should be used directly; if the S value is higher than 0.9, then the maximum fertilizer application rate of 200 kg / ha should be used directly. If the minimum fertilizer application rate is 30 kg / ha, then the minimum application rate should be adopted.

[0063] S3, based on the basic fertilization demand, corrects the preset fertilization actuator flow-response delay model and combines it with a phase lead compensation algorithm to convert the basic fertilization demand into a fertilization control signal.

[0064] Specifically, fertilization actuators, such as variable displacement pumps or valves, are modeled as a first- or second-order inertial system with pure time delay. A phase lead compensation algorithm, acting as a digital controller, is planned based on these model parameters. When the basic fertilization demand sequence is input to the controller, it outputs a sequence of control signals. For example, when the demand needs to jump from 10 units to 20 units, the controller will instantaneously output a control signal greater than 20 units, driving the actuator to reach the target flow rate more quickly, thus compensating for the inherent mechanical and fluid response delays. This fertilization control signal is typically a pulse width modulation (PWM) signal or an analog voltage of 0 to 5 volts.

[0065] In an optional embodiment, based on the basic fertilization requirement, a preset fertilization actuator flow-response delay model is corrected, and combined with a phase lead compensation algorithm, the basic fertilization requirement is converted into a fertilization control signal, including:

[0066] Based on the basic fertilization requirement, the preset flow-response delay model of the fertilization actuator is corrected; and based on the corrected flow-response delay model, a transfer function is adopted. (in The phase lead compensation algorithm of ) is as follows, where, This is the gain coefficient. and It is a time constant. The Laplace operator converts the basic fertilizer requirement into a compensated control quantity; the compensated control quantity is then converted into a duty cycle setpoint to generate a PWM signal, which serves as the fertilizer control signal.

[0067] Specifically, the calculated basic fertilization rate is converted into an electrical signal that can quickly and accurately control the hardware. Using the basic fertilization requirement of 87.8 kg / ha as the target value, the internal actuator model is calibrated in real time. The model represents the relationship between the control signal and the actual fertilizer flow rate. Calibration can adapt to changes in fertilizer properties or pipeline pressure, ensuring the correctness of the commands. To overcome the physical delay in the opening and closing of the fertilization valve, a phase lead compensation algorithm is adopted. Assume the algorithm parameters are set as follows: =1.1, =0.6s, =0.2s. When the fertilizer requirement jumps from one value to 87.8 kg / ha, the algorithm generates a momentary higher command, prompting the valve to reach the target opening more quickly, thereby reducing response lag and preventing under- or over-fertilization when the rate changes.

[0068] After processing by the compensation algorithm, an adjusted, compensated control quantity is obtained. This control quantity is a theoretical flow rate value and needs to be converted into a signal that the actuator can recognize. The actuator is typically driven by a pulse width modulation (PWM) signal. The compensated control quantity is mapped to a duty cycle value. For example, if the compensated instantaneous flow rate demand is 65% of the maximum flow rate, a PWM signal with a 65% duty cycle is generated. This high-frequency switching signal is sent to the electromagnetic controller of the fertilizer valve. By precisely controlling the energizing time ratio, accurate regulation of the fertilizer flow rate is achieved. This PWM signal is the fertilizer control signal.

[0069] S4 combines the fertilization control signal and the position correction vector to generate a fertilization execution command, which drives the fertilization actuator to perform fertilization operations at the calibrated target position.

[0070] Specifically, the control system adds a corresponding position correction vector to each sensor measurement position to obtain a queue of fertilization target positions. It continuously monitors the real-time position of the fertilization actuator provided by GNSS. When the actuator's real-time position enters within a preset trigger radius (e.g., 0.2m) of a target position in the queue, it sends a fertilization control signal bound to that target position to the actuator's drive circuit. This allows for precise, point-to-point, and quantitative fertilization at the corrected target point, with a compensated response speed.

[0071] In an optional embodiment, the fertilization control signal and the position correction vector are combined to generate a fertilization execution command, including:

[0072] Current real-time GPS location coordinates With position correction vector Perform vector addition to obtain the coordinates of the target fertilization location. ;

[0073] The target fertilization location coordinates and the PWM duty cycle value determined by the fertilization control signal are packaged into a data frame and sent as a fertilization execution command to the fertilization actuator controller.

[0074] Specifically, determine the target geographic coordinates for the fertilization operation. Obtain the current GPS location of the sensor, such as its east-west coordinates. The coordinates are 450123.4m, north-south direction. The value is 4432109.8m. Adding this to the previously calculated position correction vector, the east-west correction component ΔE is 3.55m, and the north-south correction component ΔN is 2.05m. East-west coordinates of the target fertilization location. The coordinates are 450126.95m, north-south direction. It is 4432111.85m.

[0075] The target location coordinates are bound to the generated fertilization control signal. The specific form of the fertilization control signal is a PWM duty cycle, for example, 65%. The target location coordinates 450126.95 and 4432111.85, along with the PWM duty cycle value of 65, are packaged into a standard data frame. This data frame is the fertilization execution command. This command is sent to the dedicated controller of the fertilization actuator via the vehicle communication bus. After receiving the command, the controller continuously monitors the real-time position of the agricultural machinery. Once the GPS position matches the target position in the command, the controller immediately sets the PWM output to a 65% duty cycle, driving the valve to apply fertilizer as required by the command, thereby achieving the goal of applying precise fertilizer at precise locations.

[0076] An embodiment of the adaptive fixed-point and quantitative fertilization control system provided by the present invention:

[0077] The adaptive fixed-point and quantitative fertilization control system includes a processor and a memory. The memory stores computer program instructions, which are executed by the processor to implement the adaptive fixed-point and quantitative fertilization control method described above.

[0078] The adaptive fixed-point and quantitative fertilization control system also includes other components well known to those skilled in the art, such as communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0079] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

[0080] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. An adaptive method for controlling fixed-point and quantitative fertilization, characterized in that, Includes the following steps: S1: Acquire the real-time position, speed, and heading angle of the agricultural machinery, and simultaneously collect multispectral images of the crop canopy and soil electrical conductivity data at the current location; extract crop canopy spectral indices from the crop canopy multispectral images, and construct a spatiotemporal influence weight matrix representing the spatial correlation of crop nutrient requirements based on the historical and current crop canopy spectral indices of multiple sampling points within the current location and a preset spatial neighborhood, including: Centered on the current location, set a spatial neighborhood grid of a preset size; obtain all sampling points within the neighborhood at the current time. and historical moments The crop canopy spectral index, j=1, 2, ..., T; The spatial weights of each neighboring point i and the center point are calculated using the Gaussian decay function. ,in Let i be the Euclidean distance between the neighboring point i and the center point. For spatial attenuation scale parameters; Historical moments are calculated using an exponential decay function. Relative to the current time Time weight ,in This is the time decay coefficient; Multiplying the spatial weight by the temporal weight yields the spatiotemporal influence weight of each neighborhood sampling point at different times. And combine them into a spatiotemporal influence weight matrix; S2, calculate the position correction vector to compensate for the spatial lag between the sensor measurement position and the fertilizer actuator action position based on the speed and heading angle of the agricultural machinery; Using crop canopy spectral index and soil electrical conductivity data at the current location as input features, a weighted fusion is performed using a spatiotemporal influence weight matrix to obtain a comprehensive state parameter. This includes: performing minimum-maximum normalization on all crop canopy spectral indices within the spatial neighborhood and historical periods, as well as soil electrical conductivity data at the current location, to the [0, 1] interval; and then using the spatiotemporal influence weight matrix to perform a weighted summation on all normalized crop canopy spectral indices to obtain a spatiotemporally weighted spectral index. Spatiotemporal weighted spectral index Compared with normalized soil electrical conductivity data According to the preset weighting coefficients and Perform linear summation to obtain the comprehensive state parameters. ,in ; The comprehensive state parameters are then substituted into a preset double S-shaped saturated response function to calculate the basic fertilizer requirement at the current location, including: Set the crop nutrient stress threshold for the comprehensive state parameter S. With nutrient adequacy threshold ;when At that time, the basic fertilizer requirement is the preset maximum fertilizer application rate. ;when At that time, the basic fertilizer requirement is the preset minimum fertilizer amount. ;when At that time, the basic fertilizer requirement F is calculated using the following formula: Where k is the curve steepness coefficient. In response to the midpoint value, and ; S3, based on the basic fertilization demand, corrects the preset fertilization actuator flow-response delay model, and combines it with the phase advance compensation algorithm to convert the basic fertilization demand into a fertilization control signal. S4 combines the fertilization control signal and the position correction vector to generate a fertilization execution command, which drives the fertilization actuator to perform fertilization operations at the calibrated target position.

2. The adaptive fixed-point and quantitative fertilization control method according to claim 1, characterized in that, Based on the speed and heading angle of the agricultural machinery, a position correction vector is calculated to compensate for the spatial lag between the sensor measurement position and the fertilizer actuator's operating position, including: Obtain the fixed physical distance d between the sensor and the fertilizer actuator, as well as the inherent response delay time. The real-time speed v of the agricultural machinery is compared with the response delay time. Multiplication yields the delayed displacement Add the physical distance d to the delayed displacement s to obtain the total distance D = d + s; combine this with the real-time heading angle. Calculate the correction components of the correction vector in the east-west direction. and the correction components in the north-south direction , forming a position correction vector .

3. The adaptive fixed-point and quantitative fertilization control method according to claim 1, characterized in that, Based on the basic fertilization demand, the preset flow-response delay model of the fertilization actuator is corrected, and combined with the phase lead compensation algorithm, the basic fertilization demand is converted into a fertilization control signal, including: Based on the basic fertilization requirement, the preset flow-response delay model of the fertilization actuator is corrected; and based on the corrected flow-response delay model, a transfer function is adopted. The phase lead compensation algorithm, in which, , This is the gain coefficient. and It is a time constant. The Laplace operator converts the basic fertilizer requirement into a compensated control quantity; the compensated control quantity is then converted into a duty cycle setpoint to generate a PWM signal, which serves as the fertilizer control signal.

4. The adaptive fixed-point and quantitative fertilization control method according to claim 1, characterized in that, By combining the fertilization control signal and the position correction vector, fertilization execution instructions are generated, including: Current real-time GPS location coordinates With position correction vector Perform vector addition to obtain the coordinates of the target fertilization location. ; The target fertilization location coordinates and the PWM duty cycle value determined by the fertilization control signal are packaged into a data frame and sent as a fertilization execution command to the fertilization actuator controller.

5. The adaptive fixed-point and quantitative fertilization control method according to claim 1, characterized in that, In S1, the real-time position is obtained through RTK-GNSS, the heading angle is obtained through IMU, and the velocity is calculated from the rate of change of the GNSS coordinate points over time.

6. The adaptive fixed-point and quantitative fertilization control method according to claim 1, characterized in that, In S3, the fertilizer control signal is an analog voltage of 0 to 5 volts.

7. An adaptive fixed-point and quantitative fertilization control system, characterized in that, It includes a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the adaptive fixed-point quantitative fertilization control method according to any one of claims 1-6 is implemented.

Images