A seeding operation sowing rate and depth collaborative regulation system based on forward-looking visual working condition recognition and dynamic time lag compensation

Abstract

The application relates to a sowing operation sowing amount and sowing depth collaborative regulation system based on forward-looking visual working condition recognition and dynamic time lag compensation, which comprises a main controller, a man-machine interaction module, a running speed detection module, a forward-looking visual working condition recognition module, a sowing amount regulation module, a sowing depth detection module, a sowing depth regulation module and a sowing quality feedback module, the forward-looking visual working condition recognition module is used for acquiring ground surface image information in front of a sowing unit and recognizing a sowing operation working condition category or a ground surface disturbance feature; the main controller generates a sowing depth feedforward compensation amount based on the working condition category or the ground surface disturbance feature, and feeds back and corrects the sowing depth regulation module in combination with sowing depth information output by the sowing depth detection module; and the main controller also performs dynamic time lag compensation calculation on operation parameters of the sowing amount regulation module based on running speed information and running speed change information; the application can reduce sowing depth control lag and sowing amount dynamic error under complex working conditions, and improve sowing operation quality stability.

Description

Technical Field

[0001] This invention relates to the field of agricultural tillage technology, specifically a seeding operation seeding quantity and seeding depth coordinated control system based on forward vision condition recognition and dynamic time delay compensation. Background Technology

[0002] Sowing is a crucial step in agricultural production, and the stability of the seeding rate and the consistency of the sowing depth directly affect crop emergence quality and yield. In existing sowing control systems, one type of scheme primarily uses speed-sensitive control of the sowing actuator to adjust the seeding rate; another type primarily uses sensors such as ultrasonic waves, tilt angles, pressure, or displacement to detect the sowing depth and then uses the actuator to achieve closed-loop adjustment of the sowing depth.

[0003] However, the above technical solutions still have the following problems: 1. Current seeding depth adjustments are mostly based on the current seeding depth or the seeding depth deviation that has occurred. When sudden changes occur, such as surface undulations, stubble accumulation, or soil blockages, seeding depth control has a response lag. 2. Existing speed-based seeding rate control methods mostly calculate the target rotation speed of the seeding actuator directly based on the current travel speed, without fully considering the response delay of the seeding actuator and the delay in seed delivery and soil entry. This can easily lead to dynamic errors in seeding rate under acceleration and deceleration conditions. 3. Existing missed seed detection methods mostly remain at the level of abnormal alarms, and the detection results fail to be used in reverse to correct seeding rate and seeding depth parameters, resulting in a disconnect between seeding quality monitoring and seeding rate and seeding depth control.

[0004] Therefore, it is necessary to provide a new seeding operation control system to achieve early identification of complex working conditions, time-delay compensation for dynamic errors in seeding rate, and coordinated correction of missed seeding quality feedback. Summary of the Invention

[0005] The technical problem this invention aims to solve is as follows: Existing seeding operation control systems suffer from issues such as seeding depth adjustment relying primarily on feedback control, adjustment lag under sudden surface changes, seeding rate control failing to adequately consider actuator response delays and seed delivery delays, susceptibility to dynamic errors under acceleration and deceleration conditions, and the fact that missed seeding detection results are mostly used only for abnormal alarms and not for collaborative correction of seeding rate and seeding depth parameters. This invention provides a seeding operation seeding rate and seeding depth collaborative control system based on forward-looking visual condition recognition and dynamic time delay compensation, to improve the stability and consistency of seeding operation quality under complex conditions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A seeding operation seeding quantity and seeding depth coordinated control system based on forward vision condition recognition and dynamic time delay compensation includes a main controller, a human-computer interaction module, a travel speed detection module, a forward vision condition recognition module, a seeding quantity control module, a seeding depth detection module, a seeding depth control module, and a seeding quality feedback module. The main controller is communicatively connected to the human-machine interaction module, the travel speed detection module, the forward vision condition recognition module, the sowing depth detection module, the sowing depth control module, the seeding quantity control module, and the sowing quality feedback module. The main controller is used to receive and process the information collected by each module, and generate seeding quantity control instructions and sowing depth control instructions based on preset operating parameters and control strategies. The human-computer interaction module is used to set operation parameters, control system start and stop, and display operation status. The travel speed detection module is used to acquire travel speed information in real time during the sowing operation; The forward-looking visual condition recognition module is used to acquire ground surface image information in front of the sowing unit, identify the sowing operation condition category or ground disturbance state, and send the recognition result to the main controller. The forward-looking visual condition recognition module is a deep learning model obtained by training with a training set. The training samples in the training set are ground undulation images, stubble cover images, soil clod obstacle images, furrow boundary images, and field turn images with known sowing operation condition categories or ground disturbance states. The sowing depth detection module is used to acquire sowing depth information in real time during the sowing operation and send the sowing depth information to the main controller; The sowing depth control module includes a sowing depth adjustment mechanism connected to the sowing unit; The sowing quality feedback module is used to detect missed sowing or abnormal sowing status of the sowing unit in real time, and send the detection results to the main controller; The seeding rate control module includes a seeding execution mechanism connected to the seed metering device; The main controller is used to switch different control modes according to the forward vision condition recognition results. The main controller is also used to select the sowing depth feedforward compensation amount of the corresponding control mode according to the forward vision condition recognition results. The sowing depth feedforward compensation amount is used to correct the target position or adjustment amount of the sowing depth adjustment mechanism in advance before the sowing unit reaches the surface disturbance area. The sowing quantity control parameters, sowing depth feedforward compensation amount and sowing depth feedback correction parameters are set differently under different control modes. The main controller is used to determine the comprehensive seeding depth adjustment amount based on the target seeding depth H0, the real-time seeding depth H, and the forward vision condition recognition result, wherein the difference between the real-time seeding depth H and the target seeding depth H0 is the seeding depth feedback correction amount ΔH. b ΔH, depth feedforward compensation fGenerated by the mode selection of the main controller, the main controller will provide a feedback correction amount ΔH for the broadcast depth. b With the feedforward compensation amount ΔH of the seeding depth f The summation yields the comprehensive seeding depth adjustment ΔH. c And based on the comprehensive seeding depth adjustment amount ΔH c Drive the sowing depth adjustment mechanism to achieve coordinated control of sowing depth; Based on the travel speed information, travel speed change information, and preset sowing parameters, the main controller performs dynamic time delay compensation calculation on the target operating parameters of the sowing actuator, combined with the response delay of the sowing actuator and the seed delivery delay, and outputs control signals to drive the sowing actuator to operate, so as to achieve seeding rate regulation. The main controller is also used to collaboratively correct the seeding rate control parameters and / or seeding depth control parameters based on the seeding quality feedback results.

[0007] Furthermore, the main controller is a microcontroller, and the main controller communicates with the human-machine interaction module, the travel speed detection module, the forward vision condition recognition module, the sowing depth detection module, the sowing quality feedback module, and each actuator via a communication bus, and the communication bus is a CAN bus.

[0008] Furthermore, the human-machine interaction module is a touch screen display. The human-machine interaction module is used to set parameters such as seeding rate per acre, row spacing, number of rows, target seeding depth, working mode, and alarm threshold. It is also used to display the working status, working speed, seeding quality status, and abnormal information.

[0009] Furthermore, the forward-looking visual condition recognition module includes a camera, which is installed in front of the sowing unit or the front of the frame to collect surface image information in front of the sowing unit. The condition categories or surface disturbance features identified by the forward-looking visual condition recognition module include at least one or more of the following: surface undulation, stubble cover, soil clod obstacles, furrow boundaries, or field bends.

[0010] Furthermore, the dynamic time delay compensation calculation is based at least on the travel speed information, the travel speed change rate, the response delay of the sowing execution mechanism, and the seed delivery delay to determine the target operating parameters of the sowing execution mechanism; when the target plant spacing or target seeding amount remains unchanged, the basic target rotation speed of the sowing execution mechanism increases with the increase of travel speed and decreases with the decrease of travel speed, and the main controller makes advance corrections to the basic target rotation speed based on the travel speed change rate and the total time delay.

[0011] Furthermore, the main controller determines the sowing frequency based on the current travel speed v and the target plant spacing S, and determines the basic target rotation speed n0 of the sowing actuator based on the number of seeds sown per unit rotation cycle of the seed metering device; when the travel speed is detected to be in an acceleration state, the main controller increases the compensated target rotation speed n1 based on the predicted speed; when the travel speed is detected to be in a deceleration state, the main controller decreases the compensated target rotation speed n1 based on the predicted speed.

[0012] Furthermore, the sowing depth detection module includes at least an ultrasonic sensor and a tilt sensor. The main controller fuses the ranging data from the ultrasonic sensor and the attitude data from the tilt sensor to obtain sowing depth information.

[0013] Furthermore, the sowing quality feedback module includes an infrared missed sowing detection sensor, which includes an infrared transmitter and an infrared receiver. The infrared transmitter and receiver are respectively disposed on both sides of the seed dispensing channel or seed guide tube to form an infrared detection optical path that passes through the seed falling path. When a seed passes through the infrared detection optical path, the intensity of the infrared signal received by the infrared receiver changes. The main controller generates a seed passing pulse based on the change in infrared signal intensity and determines the missed sowing or abnormal sowing status based on the seed passing pulse.

[0014] Furthermore, the sowing depth control module adopts a dual-threshold hysteresis logic control method. When the absolute value of the sowing depth deviation is greater than the first threshold, the adjustment is triggered. When the absolute value of the sowing depth deviation is less than the second threshold, the adjustment is stopped and the current state is maintained. The first threshold is greater than the second threshold.

[0015] Furthermore, the main controller is used to collaboratively correct the seeding rate control parameters, seeding depth target value, or seeding depth adjustment threshold based on the missed seeding rate, abnormal seeding frequency, or abnormal seeding location characteristics. The main controller is also used to determine the theoretical seeding time window based on the target rotation speed of the seeding execution mechanism, the number of seeds per unit angle of the seed metering device, or the target plant spacing. When no seed passing pulse is detected within the theoretical seeding time window, or when the number of detected seed passing pulses is less than the theoretical seeding quantity, it is determined to be an abnormal missed seeding.

[0016] Compared with the prior art, the present invention has at least the following beneficial effects: 1. By recognizing forward-looking visual conditions, the early prediction of seeding depth disturbances can be achieved, which can reduce seeding depth errors caused by the lag in seeding depth adjustment under sudden changes in surface conditions. 2. By introducing dynamic time delay compensation in seeding rate control, the dynamic error in seeding rate caused by the response delay of the actuator and the seed delivery delay under acceleration and deceleration conditions can be reduced; 3. By inputting the results of missed or abnormal sowing detection as sowing quality feedback into the main controller, the coordinated correction of sowing quantity parameters and sowing depth parameters can be achieved, no longer limited to abnormal alarms; 4. By integrating feedforward compensation, feedback correction, and quality feedback, the stability, consistency, and intelligence of seeding operations under complex working conditions can be improved.

[0017] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. Attached Figure Description

[0018] Figure 1 This is a diagram of the overall system architecture of the present invention; Figure 2 Flowchart for dynamic time lag compensation for broadcast volume; Figure 3 The flowchart is for the feedforward-feedback composite control of the propagation depth based on forward vision condition recognition. Figure 4 Flowchart for collaborative correction of seeding quality feedback; Figure 5 A schematic diagram showing the comparison before and after seeding depth feedforward compensation control; Figure 6 This is a schematic diagram showing the installation positions of the components of the system of the present invention on the seeder.

[0019] The components represented by the numbers in the attached diagram are as follows: 1. Main controller; 2. Human-machine interaction module; 3. Travel speed detection module; 4. Forward vision condition recognition module; 5. Sowing depth detection module; 6. Sowing quality feedback module; 7. Sowing depth control module; 8. Seeding rate control module; 9. Communication bus. Detailed Implementation

[0020] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0021] like Figure 1 and Figure 6As shown, this invention provides a seeding operation seeding quantity and seeding depth coordinated control system based on forward-looking visual condition recognition and dynamic time delay compensation. The system includes a main controller 1, a human-machine interface module 2, a travel speed detection module 3, a forward-looking visual condition recognition module 4, a seeding depth detection module 5, a seeding quality feedback module 6, a seeding depth control module 7, a seeding quantity control module 8, and a communication bus 9. The main controller 1 serves as the system's information processing and control center, receiving and processing information collected by multiple sensors, and generating seeding quantity and seeding depth control commands based on the operating parameters set by the human-machine interface module 2 and preset control strategies. This embodiment further configures a forward-looking visual condition recognition module and a seeding quality feedback module on the basis of the main controller, touch display, travel speed detection, seeding depth detection, and seeding depth control structure to achieve coordinated control of seeding quantity and seeding depth.

[0022] In one implementation, the human-machine interface module 2 is a touch screen display screen used to set parameters such as seeding rate per acre, row spacing, number of rows, target seeding depth, operating mode, and alarm threshold. It also displays the seeding operation status, operating speed, seeding depth status, missed seeding status, and abnormal information. System start-up and shutdown are controlled by the human-machine interface module 2. The touch screen display screen communicates with the main controller 1 via the communication bus 9 for parameter distribution, status feedback, and interface refresh. In addition to parameter setting and operation data display functions, the touch screen display screen is also used to set operating mode parameters and alarm threshold parameters.

[0023] In one implementation, the main controller 1 communicates with the human-machine interaction module 2, the travel speed detection module 3, the forward vision condition recognition module 4, the sowing depth detection module 5, the sowing quality feedback module 6, the sowing depth control module 7, and the seeding rate control module 8 via the communication bus 9; the communication bus 9 is preferably a CAN bus, used for data interaction between each module and the main controller 1.

[0024] In one implementation, the travel speed detection module 3 employs a radar speed measuring device, which is mounted on the seeder frame at a certain angle to the seedbed surface, and is used to output real-time travel speed information for the sowing operation. After acquiring the travel speed information, the main controller 1 not only uses it for conventional seeding rate control calculations, but also further calculates the rate of change of travel speed for dynamic time delay compensation.

[0025] In one implementation, the forward-looking visual condition recognition module 4 includes a camera, which is installed in front of the sowing unit or the front of the frame, facing the ground surface in front of the sowing unit, and is used to collect image information of the ground surface in front of the sowing unit. The main controller 1 processes the image information to identify the sowing operation condition category or ground disturbance characteristics. The operation condition category or ground disturbance characteristics include at least one or more of the following: ground undulation, stubble cover, soil clod obstacles, furrow boundaries, and field bends.

[0026] In one implementation, the main controller 1 can identify the level of surface disturbance by image grayscale features, edge features, texture features, color features or contour features, and classify the level of surface disturbance into three levels: low disturbance, medium disturbance and high disturbance; or classify the working conditions into normal sowing conditions, field turning conditions, undulating slope conditions and high stubble conditions.

[0027] As one implementation method, the recognition result of the forward vision condition recognition module 4 is not only used for display, but also serves as a direct input for subsequent depth feedforward compensation and control mode switching.

[0028] In one implementation, the seeding rate control module 8 includes a seeding execution mechanism connected to the seed metering device. The seeding execution mechanism can be a servo motor, a stepper motor, or a DC motor, preferably a servo motor. Based on travel speed information, seeding rate per acre parameters, row spacing parameters, number of rows parameters, and seeding characteristics per unit angle of the seed metering device, the main controller 1 first calculates the basic target rotational speed of the seeding execution mechanism. Based on this, it then performs dynamic time-delay compensation by combining the travel speed change rate, the response delay of the seeding execution mechanism, and the seed delivery delay to obtain the compensated target rotational speed. Finally, it outputs a control signal to drive the seeding execution mechanism, thereby reducing seeding errors under dynamic operating conditions.

[0029] As one implementation method: In order to make dynamic time delay compensation feasible, the main controller 1 calculates the rate of change of travel speed a based on the travel speed information in two consecutive sampling periods; estimates the target sowing demand at future times based on the response delay τ1 of the sowing execution mechanism and the seed delivery delay τ2; and corrects the basic target rotation speed n0 based on the target sowing demand at future times to obtain the compensated target rotation speed n1.

[0030] As one implementation method, there is a positive correlation between the target rotation speed of the seeding actuator and the traveling speed of the seeder. Let the current traveling speed be v, the target plant spacing be S, and the seed metering device discharge Z seeds per revolution. Then the basic target rotation speed n0 can be determined according to formula (1): (1) Where n0 is the basic target rotational speed of the seed metering device, in r / min; v is the traveling speed of the seeder, in m / s; S is the target plant spacing, in m; and Z is the number of seeds metered per revolution of the seed metering device. Therefore, when the target plant spacing S and the seed metering parameter Z remain constant, the greater the traveling speed v, the greater the basic target rotational speed n0; and the smaller the traveling speed v, the smaller the basic target rotational speed n0.

[0031] To further compensate for the response delay of the sowing actuator and the seed transport delay from the seed metering device to the soil entry position, the main controller 1 estimates the predicted speed v′ based on the current speed v, the rate of change of speed a, and the total time delay τ, where τ includes the response delay of the sowing actuator τ1 and the seed transport delay τ2, τ=τ1+τ2. The predicted speed v′ can be determined according to formula (2): (2) The main controller 1 then calculates the compensated target rotational speed n1 based on the predicted speed v′, which can be determined according to formula (3): (3) When the seeder accelerates, 'a' is a positive value, the predicted speed v′ is greater than the current speed v, and the compensated target rotational speed n1 is higher than the basic target rotational speed n0, thus increasing the seeding speed in advance. When the seeder decelerates, 'a' is a negative value, the predicted speed v′ is less than the current speed v, and the compensated target rotational speed n1 is lower than the basic target rotational speed n0, thus decreasing the seeding speed in advance. Through this method, the rotational speed of the seeding actuator does not simply lag behind the current speed change, but adjusts in advance according to the speed change trend to reduce the dynamic error of seeding rate under acceleration and deceleration conditions.

[0032] like Figure 2 As shown, the execution process of dynamic time delay compensation for seeding is as follows: After the system starts, the operation parameters such as seeding rate per mu, row spacing, and number of rows are set through the human-machine interaction module 2; the main controller 1 collects the speed data of the traveling speed detection module 3 in real time and calculates the speed change rate; the compensation target rotation speed of the seeding control module 8 is calculated according to the basic seeding model and the dynamic time delay compensation model; then the control signal is output to drive the seeding execution mechanism to run; at the same time, the seeding status, operation speed, operation area and abnormal status information are sent back to the human-machine interaction module 2 for display.

[0033] As one implementation method: the recognition area of ​​the forward-looking vision condition recognition module 4 and the soil entry position of the disc trencher have a forward-looking distance L. f The main controller calculates the time t it takes for the disturbance to reach the position of the disk trencher based on the travel speed v. f The calculation formula (4) is shown below: (4) The main controller 1 generates the seeding depth feedforward compensation amount ΔH based on the surface height change Δh or disturbance level obtained from forward vision recognition. f When surface protrusions, soil clods, or stubble accumulations are detected, the disc furrower tends to be raised, resulting in shallower seeding depth. In this case, a positive feedforward compensation is generated. When trenches, depressions, or soft, subsided areas are detected, the disc furrower tends to penetrate too deeply into the soil. In this case, a negative feedforward compensation is generated.

[0034] Seeding depth feedforward compensation ΔH f The calculation formula (5) can be expressed as: (5) Where, k f Δh is the feedforward compensation coefficient, and Δh is the change in ground height obtained from forward-looking visual recognition.

[0035] When the change in ground height is not calculated directly, ΔH can be determined by looking up a table based on the type of working condition or the level of disturbance. f For example, low, medium, and high disturbances correspond to different feedforward compensation amounts. To avoid over-adjustment, ΔH f Set an upper and lower limit for compensation, namely: ΔH f ∈[ΔH f min, ΔH f max).

[0036] As one implementation method, when the identification result is a high stubble cover or surface protrusion condition, the main controller 1 increases the seeding depth feedforward compensation; when the identification result is a field bend or boundary condition, the main controller 1 switches to the condition protection mode, reduces the seeding depth adjustment amplitude or reduces the adjustment frequency, so as to improve control stability.

[0037] In one implementation, the sowing depth control module 7 includes a sowing depth adjustment mechanism connected to the sowing unit. This mechanism can be an electric push rod, a hydraulic actuator, or a pneumatic actuator, preferably an electric push rod. When an electric push rod is used, the main controller 1 controls the extension and retraction of the electric push rod via a relay to adjust the height of the sowing unit.

[0038] As one implementation method, the sowing depth control module 7 preferably adopts a dual-threshold hysteresis logic control method. Adjustment is triggered when the absolute value of the overall sowing depth deviation is greater than a first threshold; adjustment stops and the current state is maintained when the absolute value of the overall sowing depth deviation is less than a second threshold, wherein the first threshold is greater than the second threshold. This method avoids frequent operation of the sowing depth adjustment mechanism due to minor disturbances under complex working conditions, thus improving control stability.

[0039] like Figure 3As shown, the feedforward-feedback composite control process for sowing depth based on forward-looking visual condition recognition is as follows: the target sowing depth and threshold parameters are set through the human-machine interaction module 2; the forward-looking visual condition recognition module 4 collects real-time images of the ground surface in front of the sowing unit and identifies the condition category or ground disturbance level; the main controller 1 generates a feedforward compensation amount for sowing depth based on the recognition results; at the same time, the main controller 1 acquires the sowing depth information output by the sowing depth detection module 5 and performs fusion calculation, compares the fused sowing depth with the target sowing depth, and obtains the feedback correction amount; the main controller 1 calculates the comprehensive control amount based on the feedforward compensation amount and the feedback correction amount, and drives the sowing depth adjustment mechanism to adjust; when the comprehensive sowing depth deviation falls within the second threshold, the current state remains unchanged.

[0040] like Figure 5 As shown, the soil penetration depth of the disc furrow opener corresponds to the seed furrow depth. Seeds fall from below the disc furrow opener into the bottom of the seed furrow. The actual sowing depth is the vertical distance between the center of the seed and the ground surface directly above it. The target sowing depth is denoted as H0, the actual sowing depth before adjustment is denoted as H1, and the actual sowing depth after adjustment is denoted as H2. Before adjustment, the sowing depth adjustment mechanism did not perform timely feedforward compensation or the compensation was insufficient, resulting in insufficient soil penetration depth of the disc furrow opener, shallow seed furrows, and seeds falling into the bottom of shallow seed furrows. At this time, the actual sowing depth H1 is less than the target sowing depth H0. After adjustment, the forward vision condition recognition module identifies the surface undulations, soil clods, or stubble disturbance in front of the sowing unit in advance. The main controller generates a sowing depth feedforward compensation amount based on the recognition results and controls the electric push rod to drive the sowing depth adjustment mechanism to make the soil penetration depth of the disc furrow opener close to the target sowing depth H0. The seeds fall into the bottom of the seed furrow opened by the disc furrow opener, so that the actual sowing depth H2 after adjustment is close to the target sowing depth H0, i.e., H2≈H0.

[0041] As one implementation method, the sowing quality feedback module 6 employs infrared missed seed detection. The infrared missed seed detection sensor is installed at the seed metering channel or seed guide tube, and includes an infrared transmitter and an infrared receiver. The infrared transmitter sends an infrared detection signal to the infrared receiver, forming a detection optical path that traverses the seed's descent path. When no seeds pass through, the infrared signal received by the infrared receiver remains within a stable range; when a seed passes through the detection optical path, the seed blocks or reflects the infrared light, causing a momentary change in the output signal of the infrared receiver. The main controller 1 performs threshold comparison and pulse shaping on this signal to generate a seed passage pulse.

[0042] The main controller 1 calculates the theoretical seeding cycle or theoretical seeding time window under the current operating conditions based on the target rotation speed of the seeding actuator, the number of holes in the seed metering disc, or the number of seeds per unit rotation angle. If no seed passing pulse is detected within the theoretical seeding time window, or if the number of detected pulses is less than the theoretical number of seeds that should pass, it is determined to be a missed seeding. If the number of detected pulses increases abnormally within a unit of time or the pulse interval is significantly less than the theoretical interval, it is determined to be a reseeding or abnormal seeding. Therefore, the seeding quality feedback module 6 does not simply issue an alarm, but feeds back the missed seeding status, number of missed seedings, and missed seeding rate obtained from infrared detection to the main controller 1 for coordinated correction of seeding quantity control parameters and / or seeding depth control parameters.

[0043] Specifically, when the missed seeding rate exceeds a set threshold and drastic changes in seeding velocity are detected simultaneously, the main controller 1 prioritizes increasing the dynamic time-delay compensation for seeding rate control. When the missed seeding rate exceeds a set threshold and a high level of surface disturbance is detected simultaneously, the main controller 1 prioritizes correcting the target seeding depth, the seeding depth feedforward compensation, or the seeding depth adjustment threshold. When missed seeding anomalies persist, the main controller 1 can output alarm information through the human-machine interface module 2 and prompt the operator to check. In this way, the missed seeding detection results are no longer only used for alarms but are further involved in the coordinated correction of seeding rate and seeding depth parameters.

[0044] As one implementation method, the seeding depth feedforward compensation amount ΔH f This is used to compensate for the impact of changes in ground height or ground disturbance that the sowing unit will encounter on the actual sowing depth. The forward vision condition recognition module 4 is installed in front of the sowing unit, and its recognition area is a certain forward distance L from the furrowing and soil entry position. f The main controller 1 calculates the time it takes for the ground disturbance to reach the seeding unit based on the travel speed v, and outputs a seeding depth adjustment command in advance before the seeding unit reaches the disturbance area.

[0045] When soil clods, stubble piles, or surface protrusions are detected ahead, without feedforward compensation, the seeding unit is easily lifted when passing through this area, causing the actual seeding depth to be less than the target seeding depth H0. At this time, the main controller 1 generates a positive seeding depth feedforward compensation amount ΔH. f The seeding depth adjustment mechanism is driven in advance to press down or lower the seeding unit, so that the furrow opener remains close to the target seeding depth when passing through the disturbed area.

[0046] When a trench, depression, or soft, sunken area is detected ahead, the sowing unit is prone to sinking if no feedforward compensation is performed, causing the actual sowing depth to exceed the target sowing depth H0. At this time, the main controller 1 generates a reverse sowing depth feedforward compensation amount ΔH. f The sowing depth adjustment mechanism can be activated in advance to raise or reduce the sowing unit's depth in the soil to avoid sowing too deep.

[0047] In the feedback control section, the main controller 1 compares the real-time sowing depth H detected by the sowing depth detection module 5 with the target sowing depth H0 to obtain the sowing depth feedback correction amount ΔH. b In the feedforward control section, the main controller 1 obtains the seeding depth feedforward compensation amount ΔH based on the operating condition category or disturbance level identified by forward vision. f Finally, the two are combined to obtain the comprehensive seeding depth adjustment amount ΔH. c Comprehensive seeding depth adjustment ΔH c This can be represented as ΔH c =ΔH b +ΔH f Alternatively, it can be obtained by combining different weights assigned according to different working conditions. The main controller 1 adjusts the overall seeding depth ΔH. c The system controls the movement of electric push rods, hydraulic cylinders, or pneumatic actuators to achieve a composite control system that combines advance compensation for seeding depth with real-time feedback correction.

[0048] like Figure 4 As shown, the sowing quality feedback collaborative correction process is as follows: the sowing quality feedback module 6 outputs the missed sowing rate, the number of abnormal sowings, or the location of abnormal sowings in real time; the main controller 1 combines the travel speed change information, the forward vision condition recognition results, and the sowing depth feedback information to classify and judge the cause of the abnormality; if the abnormality is mainly caused by dynamic speed changes, the sowing rate control parameters are corrected; if the abnormality is mainly caused by surface disturbance or sowing depth deviation, the sowing depth target value, sowing depth feedforward compensation amount, or sowing depth threshold parameter is corrected; the corrected parameters are reapplied to the sowing rate control module 8 and the sowing depth control module 7 to achieve quality feedback collaborative control.

[0049] As one implementation, the main controller 1 can also switch between different control modes based on the forward-looking visual condition recognition results. The control modes include at least one or more of the following: normal sowing mode, complex surface mode, field turning mode, and high stubble mode.

[0050] Under normal sowing mode, the main controller 1 executes conventional dynamic time-delay compensation and sowing depth feedforward-feedback composite control. Under complex surface mode, the weight of sowing depth feedforward compensation is increased. Under field turning mode, the amplitude of dynamic adjustment of sowing rate is reduced and the frequency of action of sowing depth actuator is limited. Under high stubble mode, the weight of missed sowing quality feedback and the level of abnormal alarm are increased. Thus, the sowing rate control parameters, sowing depth feedforward compensation amount, and sowing depth feedback correction parameters are different under different operating conditions, thereby achieving multi-mode collaborative control.

[0051] The workflow of this invention is as follows: First, the target sowing depth and threshold parameters are set through the human-computer interaction module 2; the forward vision condition recognition module 4 collects real-time images of the ground surface in front of the sowing unit and identifies the condition category or ground disturbance level; the main controller 1 generates a sowing depth feedforward compensation amount based on the recognition results; simultaneously, the main controller 1 acquires the sowing depth information output by the sowing depth detection module 5 and performs fusion calculation, compares the fused sowing depth with the target sowing depth, and obtains a feedback correction amount; the main controller 1 calculates the comprehensive control amount based on the feedforward compensation amount and the feedback correction amount, and drives the sowing depth adjustment mechanism to adjust; when the comprehensive sowing depth deviation falls within the second threshold, the current state remains unchanged.

[0052] The sowing quality feedback module 6 outputs real-time information on missed sowing rate, number of abnormal sowings, or location of abnormal sowings; the main controller 1 combines information on changes in travel speed, forward vision condition recognition results, and sowing depth feedback information to classify and judge the causes of abnormalities; if the abnormality is mainly caused by dynamic speed changes, the sowing rate control parameters are corrected; if the abnormality is mainly caused by surface disturbance or sowing depth deviation, the sowing depth target value, sowing depth feedforward compensation amount, or sowing depth threshold parameter is corrected; the corrected parameters are then applied back to the sowing rate control module 8 and the sowing depth control module 7 to achieve quality feedback collaborative control.

[0053] The above description provides examples of the preferred embodiments of the present invention. Parts not detailed herein are common knowledge to those skilled in the art. The scope of protection of the present invention is determined by the claims. Any equivalent modifications based on the technical teachings of the present invention are also within the scope of protection of the present invention.

Claims

1. A seeding operation seeding quantity and seeding depth coordinated control system based on forward-looking visual condition recognition and dynamic time delay compensation, characterized in that, It includes a main controller, a human-machine interaction module, a travel speed detection module, a forward vision condition recognition module, a seeding rate control module, a seeding depth detection module, a seeding depth control module, and a seeding quality feedback module; The main controller is communicatively connected to the human-machine interaction module, the travel speed detection module, the forward vision condition recognition module, the sowing depth detection module, the sowing depth control module, the seeding quantity control module, and the sowing quality feedback module. The main controller is used to receive and process the information collected by each module, and generate seeding quantity control instructions and sowing depth control instructions based on preset operating parameters and control strategies. The human-computer interaction module is used to set operation parameters, control system start and stop, and display operation status. The travel speed detection module is used to acquire travel speed information in real time during the sowing operation; The forward-looking visual condition recognition module is used to acquire ground surface image information in front of the sowing unit, identify the sowing operation condition category or ground disturbance state, and send the recognition result to the main controller. The forward-looking visual condition recognition module is a deep learning model obtained by training with a training set. The training samples in the training set are ground undulation images, stubble cover images, soil clod obstacle images, furrow boundary images, and field turn images with known sowing operation condition categories or ground disturbance states. The sowing depth detection module is used to acquire sowing depth information in real time during the sowing operation and send the sowing depth information to the main controller; The sowing depth control module includes a sowing depth adjustment mechanism connected to the sowing unit; The sowing quality feedback module is used to detect missed sowing or abnormal sowing status of the sowing unit in real time, and send the detection results to the main controller. The seeding rate control module includes a seeding execution mechanism connected to the seed metering device; The main controller is used to switch different control modes according to the forward vision condition recognition results. The main controller is also used to select the sowing depth feedforward compensation amount of the corresponding control mode according to the forward vision condition recognition results. The sowing depth feedforward compensation amount is used to correct the target position or adjustment amount of the sowing depth adjustment mechanism in advance before the sowing unit reaches the surface disturbance area. The sowing quantity control parameters, sowing depth feedforward compensation amount and sowing depth feedback correction parameters are set differently under different control modes. The main controller is used to determine the comprehensive seeding depth adjustment amount based on the target seeding depth H0, the real-time seeding depth H, and the forward vision condition recognition result, wherein the difference between the real-time seeding depth H and the target seeding depth H0 is the seeding depth feedback correction amount ΔH. b ΔH, depth feedforward compensation f Generated by the mode selection of the main controller, the main controller will provide a feedback correction amount ΔH for the broadcast depth. b With the feedforward compensation amount ΔH of the seeding depth f The summation yields the comprehensive seeding depth adjustment ΔH. c And based on the comprehensive seeding depth adjustment amount ΔH c Drive the sowing depth adjustment mechanism to achieve coordinated control of sowing depth; Based on the travel speed information, travel speed change information, and preset sowing parameters, the main controller performs dynamic time delay compensation calculation on the target operating parameters of the sowing actuator, combined with the response delay of the sowing actuator and the seed delivery delay, and outputs control signals to drive the sowing actuator to operate, so as to achieve seeding rate regulation. The main controller is also used to collaboratively correct the seeding rate control parameters and / or seeding depth control parameters based on the seeding quality feedback results.

2. The seeding rate and depth coordinated control system for seeding operations according to claim 1, characterized in that, The main controller is a microcontroller. The main controller communicates with the human-machine interaction module, the travel speed detection module, the forward vision condition recognition module, the sowing depth detection module, the sowing quality feedback module, and each actuator through a communication bus, which is a CAN bus.

3. The seeding rate and depth coordinated control system for seeding operations according to claim 1, characterized in that, The human-machine interaction module is a touch screen display. The human-machine interaction module is used to set parameters such as seeding rate per acre, row spacing, number of rows, target seeding depth, working mode, and alarm threshold. It is also used to display the working status, working speed, seeding quality status, and abnormal information.

4. The seeding rate and depth coordinated control system for seeding operations according to claim 1, characterized in that, The forward-looking visual condition recognition module includes a camera, which is installed in front of the sowing unit or the front of the frame to collect ground surface image information in front of the sowing unit. The condition categories or ground disturbance features identified by the forward-looking visual condition recognition module include at least one or more of the following: ground undulation, stubble cover, soil clod obstacles, furrow boundaries, or field bends.

5. The seeding rate and depth coordinated control system according to claim 1, characterized in that, The dynamic time delay compensation calculation is based at least on the travel speed information, the rate of change of travel speed, the response delay of the sowing execution mechanism, and the seed delivery delay to determine the target operating parameters of the sowing execution mechanism; when the target plant spacing or target seeding amount remains unchanged, the basic target rotation speed of the sowing execution mechanism increases with the increase of travel speed and decreases with the decrease of travel speed, and the main controller makes advance corrections to the basic target rotation speed based on the rate of change of travel speed and the total time delay.

6. The seeding rate and depth coordinated control system according to claim 5, characterized in that, The main controller determines the sowing frequency based on the current travel speed v and the target plant spacing S, and determines the basic target rotation speed n0 of the sowing actuator based on the number of seeds sown per unit rotation cycle of the seed metering device; when the travel speed is detected to be in an acceleration state, the main controller increases the compensated target rotation speed n1 based on the predicted speed; when the travel speed is detected to be in a deceleration state, the main controller decreases the compensated target rotation speed n1 based on the predicted speed.

7. The seeding rate and depth coordinated control system for seeding operations according to claim 1, characterized in that, The sowing depth detection module includes at least an ultrasonic sensor and a tilt sensor. The main controller fuses the ranging data from the ultrasonic sensor and the attitude data from the tilt sensor to obtain sowing depth information.

8. The seeding rate and depth coordinated control system according to claim 1, characterized in that, The sowing quality feedback module includes an infrared missed sowing detection sensor, which comprises an infrared transmitter and an infrared receiver. The infrared transmitter and receiver are respectively disposed on both sides of the seed dispensing channel or seed guide tube to form an infrared detection optical path that passes through the seed falling path. When a seed passes through the infrared detection optical path, the intensity of the infrared signal received by the infrared receiver changes. The main controller generates a seed passing pulse based on the change in infrared signal intensity and determines the missed sowing or abnormal sowing status based on the seed passing pulse.

9. The seeding rate and depth coordinated control system for seeding operations according to claim 1, characterized in that, The sowing depth control module adopts a dual-threshold hysteresis logic control method. When the absolute value of the sowing depth deviation is greater than the first threshold, the adjustment is triggered. When the absolute value of the sowing depth deviation is less than the second threshold, the adjustment is stopped and the current state is maintained. The first threshold is greater than the second threshold.

10. The seeding rate and depth coordinated control system for seeding operations according to claim 1, characterized in that, The main controller is used to collaboratively correct the seeding rate control parameters, seeding depth target value, or seeding depth adjustment threshold based on the characteristics of the missed seeding rate, abnormal seeding frequency, or abnormal seeding location. The main controller is used to determine the theoretical seeding time window based on the target rotation speed of the seeding execution mechanism, the number of seeds per unit angle of the seed metering device, or the target plant spacing. When no seed passing pulse is detected within the theoretical seeding time window, or when the number of detected seed passing pulses is less than the theoretical seeding quantity, it is determined to be an abnormal missed seeding.

Images