Cleaning system, method, electronic equipment and storage medium of harvester

Abstract

This invention provides a cleaning system, method, electronic device, and storage medium for a harvester, belonging to the field of agricultural machinery technology. It includes a posture detection module, a control module, and an execution module. The posture detection module acquires the harvester's body posture data and sends it to the control module. The control module predicts the posture trend based on the body posture data, obtains the prediction result, generates control commands based on the prediction result, and sends the control commands to the execution module. The execution module drives the harvester's cleaning screen to move based on the control commands. This invention, by predicting the harvester's body posture data, anticipates upcoming fluctuations in the machine body and drives the cleaning screen to actively displace, effectively solving the problem of lateral material shift and accumulation on the screen surface caused by response lag, maintaining a stable material flow distribution, and improving screen surface utilization and cleaning efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural machinery, in particular to a cleaning system and method of a harvester, an electronic device and a storage medium. BACKGROUND

[0002] The cleaning system of a combine harvester is an important device for screening and separating the threshed material during the agricultural harvesting process, thereby reducing the impurity content and loss rate of the grains. When the combine harvester is working in the field, the ground undulations will cause the machine body to tilt, thereby causing the screen surface of the cleaning system to tilt, and the threshed material on the screen surface to slide to one side.

[0003] Currently, the screen surface amplitude parameter of the cleaning system is usually adjusted by relying on monitoring feedback through sensors.

[0004] However, this method has a response lag when facing dynamic undulations of the ground, which causes the threshed material to be laterally offset and accumulated on the cleaning screen, and cannot maintain a stable material flow distribution, thereby causing low screen utilization and low cleaning efficiency. SUMMARY

[0005] The present application provides a cleaning system and method of a harvester, an electronic device and a storage medium, to solve the defects of the prior art adjustment method based on monitoring feedback, which has a response lag when facing dynamic undulations of the ground, causing the threshed material to be laterally offset and accumulated, and reducing the cleaning efficiency. The present application realizes the early prediction of the machine body undulation trend and the active displacement adjustment of the cleaning screen, thereby eliminating the response lag, maintaining uniform distribution of the material on the screen, and improving the cleaning efficiency.

[0006] The present application provides a cleaning system of a harvester, a posture detection module, a control module and an execution module; The posture detection module obtains the machine body posture data of the harvester and sends the machine body posture data to the control module; The control module performs posture trend prediction based on the machine body posture data, obtains a posture trend prediction result, generates a control instruction based on the posture trend prediction result, and sends the control instruction to the execution module; The execution module drives the cleaning screen of the harvester to move based on the control instruction.

[0007] According to the cleaning system of the harvester provided by the present application, the posture trend prediction based on the machine body posture data obtains a posture trend prediction result, which includes: Smooth the machine body posture data to obtain smoothed posture data; The smoothed attitude data is input into the attitude trend prediction model to obtain the attitude trend prediction result output by the attitude trend prediction model; The posture trend prediction model is trained based on historical posture sequence samples.

[0008] According to the cleaning system for a harvester provided by the present invention, the attitude trend prediction model is obtained based on the following training steps: Collect historical attitude sequence samples, which include the harvester's roll angle sequence, pitch angle sequence, angular rate sequence, and acceleration data during historical operating periods; Determine the future attitude label corresponding to the historical attitude sequence sample, wherein the future attitude label is the actual tilt angle of the harvester at a preset time interval after the historical attitude sequence sample collection time; The training process is executed iteratively until the preset termination condition is met. The training process includes: Input the historical attitude sequence samples into the attitude trend prediction model to obtain the predicted tilt angle output by the attitude trend prediction model; Calculate the loss between the predicted tilt angle and the future attitude label; The parameters of the attitude trend prediction model are updated based on the loss.

[0009] According to a harvester cleaning system provided by the present invention, the step of smoothing the machine body attitude data to obtain smooth attitude data includes: The attitude state equation of the harvester is constructed, and the attitude state equation is established based on the angular rate in the machine attitude data; The current angular rate and the attitude estimate from the previous preset time are input into the attitude state equation to obtain the attitude prediction value at the current time output by the attitude state equation. Calculate the target correction factor, which represents the confidence weight relationship between the current attitude prediction value and the current fuselage attitude data; The target correction factor is used to correct the attitude prediction value and the fuselage attitude data to obtain the smoothed attitude data.

[0010] According to the present invention, a cleaning system for a harvester generates control commands based on the posture trend prediction results, including: Obtain a cleaning performance table, which includes the correspondence between attitude angle ranges and lateral adjustment amounts; If the attitude trend prediction result exceeds the preset adjustment dead zone range, then the target lateral adjustment amount corresponding to the attitude trend prediction result is found according to the cleaning performance table, and the control command is generated according to the target lateral adjustment amount.

[0011] According to the present invention, a cleaning system for a harvester further includes a measurement feedback module. The measurement feedback module obtains the actual displacement of the cleaning screen and sends the actual displacement to the control module; The control module modifies the control command based on the actual displacement.

[0012] According to the present invention, a cleaning system for a harvester includes an execution module comprising an electric push rod, one end of which is connected to the frame of the harvester, and the other end of which is connected to the cleaning screen.

[0013] The present invention also provides a cleaning method for a harvester, comprising the following steps: Acquire the harvester's body attitude data; Based on the fuselage attitude data, attitude trend prediction is performed to obtain attitude trend prediction results, and control commands are generated based on the attitude trend prediction results. The control commands are sent to the execution module of the harvester.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the cleaning method of any of the harvesters described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the cleaning method of the harvester as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the cleaning method of any of the harvesters described above.

[0017] The harvester cleaning system, method, electronic equipment, and storage medium provided by this invention predict the trend of the harvester's body posture data, anticipate the upcoming fluctuations of the machine body, and drive the cleaning screen to actively displace. This effectively solves the problem of lateral displacement and accumulation of material on the screen surface caused by response lag, maintains a stable material flow distribution, and improves the utilization rate of the screen surface and cleaning efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is one of the schematic diagrams of the cleaning system for the harvester provided by the present invention.

[0020] Figure 2 This is a schematic diagram of the process for predicting attitude trends based on fuselage attitude data provided by the present invention.

[0021] Figure 3 This is a flowchart illustrating the training steps of the posture trend prediction model provided by the present invention.

[0022] Figure 4 This is a schematic diagram of the process for smoothing fuselage attitude data provided by the present invention.

[0023] Figure 5 This is a flowchart illustrating the cleaning method for the harvester provided by the present invention.

[0024] Figure 6 This is the second schematic diagram of the architecture of the cleaning system for the harvester provided by the present invention.

[0025] Figure 7 This is a schematic diagram of the installation location of the detection module provided by the present invention.

[0026] Figure 8 This is the third schematic diagram of the architecture of the cleaning system for the harvester provided by the present invention.

[0027] Figure 9 This is the fourth schematic diagram of the architecture of the cleaning system for the harvester provided by the present invention.

[0028] Figure 10 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0029] Figure label: 10 three-axis gyroscope sensor; 20 upper screen box component; 30 drive component; 40 transverse vibration component. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this 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 this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0031] It should be noted that, in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0032] The terms "upper," "lower," etc., indicating orientation or positional relationships are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Unless otherwise expressly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication between two elements. For those skilled in the art, the specific meaning of the above terms in the invention can be understood according to the specific circumstances.

[0033] The terms "first," "second," etc., used in this invention are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more.

[0034] To facilitate a full understanding of the technical solution of this application, the following content is hereby introduced: When combine harvesters are operating, undulating ground in the field can cause the machine to tilt, resulting in lateral displacement and accumulation of threshed material on the cleaning screen. This leads to low screen utilization, reduced cleaning efficiency, and increased grain loss and impurity content. Existing cleaning systems often use fixed screen structures or can only adjust screen amplitude parameters, lacking the ability to predict ground undulations. Furthermore, adjustments based on real-time monitoring feedback suffer from response lag. Some improvements achieve uniform material distribution by adding diversion mechanisms or return conveyors, but these increase mechanical complexity and material transport time, and cannot adapt to dynamically undulating operating scenarios, failing to fundamentally solve the problem of uneven material distribution on the screen caused by ground undulations.

[0035] For current applications in this field, relevant technical solutions include a three-dimensional dynamic leveling system. This system uses sensors to perceive the machine's posture in real time and actively adjusts the horizontal state of the cleaning screen. Even on a certain side slope, it maintains a level screen surface, eliminating uneven material distribution caused by slope accumulation and maintaining stable cleaning efficiency. Its three-dimensional material conveying system employs an inclined floating screen design, transporting accumulated material from both sides to the center, forming a centrally concentrated and evenly diffused material flow. This solves the saddle-shaped distribution problem caused by traditional axial flow rollers and improves screen surface utilization. Simultaneously, it utilizes pre-cleaning and main cleaning in synergy. A removable pre-cleaning plate pre-separates materials, reducing the load on the upper screen. Combined with a turbine cleaner and blower, it achieves graded cleaning.

[0036] Another related technical solution is the Self-Leveling Sieve (SLS), which uses tilt sensors and electric actuators to level in real time, maintaining the overall level of the cleaning shoe (including the grain pan, pre-screen, and upper and lower screens) under a certain side slope, eliminating material accumulation and maintaining uniform airflow penetration. This system can employ independent leveling designs or overall leveling adjustments to adapt to the load requirements of different machine models. Its cascade cleaning process separates light impurities and short straw by setting a steep-slope pre-screen, reducing the load on the main screen; the grain pan, pre-screen, and upper and lower screens are driven independently, forming a cascaded material flow of projection, suspension, and grading, extending the screening path and improving airflow utilization. Furthermore, by increasing the lateral amplitude of the screen surface, the material piled on the sloping side is guided to the center, offsetting the gravity offset of the side slope and solving the problem of single-sided accumulation on the traditional screen surface.

[0037] However, the aforementioned existing technical solutions mainly rely on passive adjustment after real-time detection. When facing continuously changing undulating terrain, the adjustment action often lags behind the change in the aircraft's attitude and cannot intervene in advance. Therefore, when dealing with highly dynamic and complex terrain, there is still room for improvement in its cleaning effect.

[0038] Therefore, the present invention provides a cleaning system, method, electronic device and storage medium for a harvester, thereby eliminating response lag, maintaining uniform material distribution on the screen surface and improving cleaning efficiency.

[0039] The following is combined with Figures 1-10 The present invention describes the cleaning system, method, electronic equipment, and storage medium for harvesters provided by the present invention.

[0040] Figure 1 This is one of the schematic diagrams of the cleaning system for the harvester provided by the present invention, such as... Figure 1 As shown, as an optional embodiment, the cleaning system of the harvester mainly includes, but is not limited to, the attitude detection module 110, the control module 120, and the execution module 130.

[0041] The attitude detection module 110 acquires the harvester's body attitude data and sends the body attitude data to the control module 120.

[0042] The attitude detection module 110 refers to a hardware unit used to sense the spatial attitude information of the harvester. For example, the attitude detection module 110 can be a sensor assembly installed on the frame of the harvester. Specifically, it can be a micro-electro-mechanical system (MEMS) inertial measurement unit that integrates an inertial measurement unit (IMU) and a triaxial accelerometer.

[0043] Machine attitude data refers to data reflecting the tilt and motion state of the harvester relative to the horizontal plane or a preset coordinate system during operation. For example, machine attitude data may include the harvester's roll angle, pitch angle, three-axis angular rate, and three-axis acceleration data.

[0044] The harvester's body attitude data can be obtained through high-frequency sampling. For example, the attitude detection module 110 can collect the aforementioned body attitude data in real time at a sampling frequency greater than or equal to 100 Hz.

[0045] The fuselage attitude data can be transmitted to the control module 120 via wired communication or vehicle network protocol. For example, the output signal of the attitude detection module 110 can be transmitted to the control module 120 via the Controller Area Network (CAN) bus interface.

[0046] The control module 120 performs attitude trend prediction based on the fuselage attitude data, obtains the attitude trend prediction result, generates control commands based on the attitude trend prediction result, and sends the control commands to the execution module 130.

[0047] Attitude trend prediction refers to the process of analyzing current and historical data to predict the state of the machine in the near future. For example, attitude trend prediction can use the prediction algorithm built into the control module 120 to calculate the tilt changes that the harvester will undergo in the future within a preset time period.

[0048] Attitude trend prediction results refer to the quantitative values ​​of the fuselage attitude at future moments output by the trend prediction algorithm. For example, the attitude trend prediction results can be the predicted lateral tilt angle or pitch angle value that the fuselage will appear at in the future.

[0049] Control commands refer to signals issued by control module 120 to instruct execution module 130 to perform actions. For example, control commands may be electrical signals containing a target extension amount or direction of movement, used to control the electric push rod of execution module 130 to extend or retract by a specific length.

[0050] The execution module 130 refers to the mechanical drive mechanism used to adjust the position of the cleaning screen. Control commands can be sent to the execution module 130 via vehicle bus communication. For example, the control module 120 can communicate with the electric actuator via the controller area network bus to send the calculated target extension command to the electric actuator's controller.

[0051] The execution module 130 drives the cleaning screen of the harvester to move based on control commands.

[0052] A cleaning screen is a screening device used in a harvester to separate and clean the threshed mixture. For example, a cleaning screen can be a vibrating screen assembly that includes an upper screen, a lower screen, and a pre-screening screen. It is installed on the fifth arm of the harvester chassis and can move relative to the machine frame.

[0053] The movement of the cleaning screen driving the harvester can be achieved through mechanical transmission. For example, the electric push rod in the execution module 130 extends or retracts according to the received control command, driving the fifth arm and the cleaning screen body rigidly connected to the fifth arm to move laterally in a direction perpendicular to the forward direction of the harvester. For example, dynamic compensation adjustment can be performed within a range of ±60 mm.

[0054] Considering that existing technologies often rely on real-time monitoring and feedback when facing ground undulations, which results in response lag and material accumulation on one side of the screen surface, affecting the cleaning effect, this invention introduces an attitude trend prediction mechanism to predict the future tilt state of the machine body in advance and control the execution module to actively adjust the lateral position of the cleaning screen. This enables advanced compensation to offset the gravitational offset caused by ground undulations, ensuring that the extruded material is always evenly distributed across the entire screen surface, significantly improving harvesting efficiency and cleaning quality.

[0055] The harvester cleaning system provided by this invention predicts the trend of the harvester's body posture data in advance, anticipates the upcoming fluctuations of the machine body, and drives the cleaning screen to actively move. This effectively solves the problem of lateral displacement and accumulation of material on the screen surface caused by response lag, maintains a stable material flow distribution, and improves the utilization rate of the screen surface and the cleaning efficiency.

[0056] As an optional implementation, this system has been proven to perform excellently in complex hilly terrain. For example, through real-time adjustment of the electric actuator, the screen surface can be kept relatively horizontal (or have an equivalent horizontal cleaning effect) even under a 20% side slope, eliminating uneven material distribution caused by slope accumulation. Compared with traditional technologies, this system can offset the gravitational shift caused by approximately 25% side slope, solving the problem of one-sided accumulation on traditional screen surfaces. Through this proactive trend prediction and compensation, harvesting efficiency can be improved by more than 15%, while significantly reducing grain loss rate and impurity content.

[0057] Figure 2 This is a schematic diagram of the process for predicting attitude trends based on fuselage attitude data provided by the present invention, as shown below. Figure 2 As shown, as another optional embodiment provided by the present invention, attitude trend prediction is performed based on fuselage attitude data to obtain attitude trend prediction results, including but not limited to the following steps: Step 210: Smooth the fuselage attitude data to obtain smoothed attitude data.

[0058] Smoothing refers to the data processing procedure that removes noise interference from data and extracts the main trend of change. For example, smoothing can be achieved by using Kalman filtering or low-pass filtering algorithms to filter out high-frequency vibration noise in the original sensor signal.

[0059] Smooth attitude data refers to a data sequence that more accurately reflects the aircraft's motion state after noise reduction processing. For example, smooth attitude data can be stable values ​​of roll angle, pitch angle, and angular rate after Kalman filtering.

[0060] Step 220: Input the smoothed attitude data into the attitude trend prediction model to obtain the attitude trend prediction result output by the attitude trend prediction model. The attitude trend prediction model is trained based on historical attitude sequence samples.

[0061] Attitude trend prediction model refers to an algorithmic model that can infer future states based on current and past states. For example, attitude trend prediction model can be a state prediction equation based on machine learning algorithms or angular rate integral calculation, which can map the relationship between the current attitude and the attitude in the short term.

[0062] As an alternative implementation, the attitude trend prediction model can be constructed using various algorithmic architectures. For example, the attitude trend prediction model can be a time series prediction model based on a recurrent neural network (RNN), or its variant, a long short-term memory (LSTM) model. These models are adept at processing time-dependent sequential data and can effectively capture the dynamic patterns of aircraft attitude changes over time.

[0063] Furthermore, the attitude trend prediction model can also be a Support Vector Regression (SVR) model, which makes predictions by learning the nonlinear relationships in historical attitude data. In embedded systems with limited computing resources, this attitude trend prediction model can also be based on the state estimation equation of Kalman Filter (KF) or Extended Kalman Filter (EKF), which establishes a state-space model of the system and uses the estimated value at the previous time step and the observed value at the current time step to recursively predict the system state at the next time step.

[0064] Historical attitude sequence samples refer to a set of attitude data collected and stored during past operating periods. For example, historical attitude sequence samples may include roll angle sequences, pitch angle sequences, angular rate sequences and acceleration data recorded by harvesters when operating under different terrain conditions.

[0065] Considering that the raw fuselage attitude data is often mixed with noise from engine vibration and road bumps, if it is used directly for prediction, the results may be inaccurate. Moreover, simple real-time feedback cannot overcome the system response delay. Therefore, this invention first smooths the data and then uses a prediction model trained on historical data to predict trends. This can effectively filter out interference signals and accurately predict the fuselage tilt status at future moments, providing a reliable basis for advance position compensation of the cleaning screen.

[0066] The harvester cleaning system provided by this invention smooths the machine posture data and uses a posture trend prediction model trained based on historical posture sequence samples to make predictions. This effectively filters out high-frequency noise interference caused by engine vibration or road bumps during harvester operation. At the same time, by utilizing the model's ability to learn from historical data patterns, it more accurately captures the dynamic change trend of the machine posture, thereby significantly improving the accuracy and robustness of the posture trend prediction results and providing a high-quality data foundation for subsequent precise control.

[0067] Figure 3This is a flowchart illustrating the training steps of the posture trend prediction model provided by the present invention, as shown below. Figure 3 As shown, as another optional embodiment provided by the present invention, the attitude trend prediction model is obtained based on the following training steps: Step 310: Collect historical attitude sequence samples, which include the harvester's roll angle sequence, pitch angle sequence, angular rate sequence, and acceleration data during historical operating periods.

[0068] Historical operation period refers to the time period during which a harvester actually performed field harvesting tasks in the past. For example, historical operation period can be a record period of continuous operation of a harvester in different terrains (such as plains, hills or slopes).

[0069] A roll angle sequence refers to a set of angle data of a harvester rotating around its longitudinal axis arranged in chronological order. For example, a roll angle sequence could be a list of left and right tilt angle values ​​of the harvester recorded every 10 milliseconds.

[0070] A pitch angle sequence refers to a set of angle data of a harvester rotating around its horizontal axis arranged in chronological order. For example, a pitch angle sequence can be time-series data of the forward and backward pitch angle values ​​of the harvester continuously collected.

[0071] An angular rate sequence refers to the data on the changes in the angular velocity of a harvester rotating around its axes. For example, an angular rate sequence can be a data stream of the three-axis angular velocities output by a gyroscope as a function of time.

[0072] Acceleration data refers to the linear acceleration information of a harvester in three-dimensional space. For example, acceleration data can be values ​​collected by a triaxial accelerometer that reflect changes in the force and motion state of the harvester.

[0073] Step 320: Determine the future attitude label corresponding to the historical attitude sequence sample. The future attitude label is the actual tilt angle of the harvester at a preset time interval after the historical attitude sequence sample collection time.

[0074] The future attitude label refers to the actual tilt state data of the fuselage, which serves as the ground truth for the training target. For example, the future attitude label could be the actual lateral tilt angle of the harvester 2 seconds after the current moment.

[0075] Step 330: Iteratively execute the training process until the preset termination condition is met. The training process includes: inputting historical posture sequence samples into the posture trend prediction model, obtaining the predicted tilt angle output by the posture trend prediction model; calculating the loss between the predicted tilt angle and the future posture label; and updating the parameters of the posture trend prediction model based on the loss.

[0076] The preset termination condition refers to the criterion for determining when the training of the attitude trend prediction model ends. For example, the preset termination condition may be that the value of the loss function converges to below a preset threshold, or that a preset number of iterations is reached.

[0077] Specifically, in each iteration, the loss between the predicted tilt angle and the future attitude label can be evaluated using loss functions such as mean squared error or cross-entropy. The parameters of the attitude trend prediction model can be updated based on the loss by adjusting the weights within the model using gradient descent or backpropagation algorithms, thereby continuously optimizing the prediction accuracy of the attitude trend prediction model.

[0078] The harvester cleaning system provided by this invention uses historical attitude sequence samples containing multi-dimensional data such as roll angle, pitch angle, angular rate and acceleration to supervise the training of the model, and uses the actual tilt angle at future moments as labels for iterative optimization. This enables the attitude trend prediction model to learn the deep rules and temporal characteristics of the aircraft's attitude changes under complex terrain, thereby significantly improving the realism and accuracy of the model's prediction of future tilt states, and ensuring that it can give highly consistent prediction results in actual operation.

[0079] Figure 4 This is a flowchart illustrating the smoothing process for fuselage attitude data provided by the present invention, as shown below. Figure 4 As shown, as another optional embodiment provided by the present invention, the fuselage attitude data is smoothed to obtain smoothed attitude data, including but not limited to the following steps: Step 410: Construct the attitude state equation of the harvester. The attitude state equation is established based on the angular rate in the machine attitude data.

[0080] The attitude state equation refers to the mathematical relationship that describes the evolution of the harvester's attitude over time. For example, the attitude state equation can be a linear or nonlinear equation based on the principle of rigid body kinematics, used to deduce the current state based on the state at the previous moment and the current input.

[0081] Specifically, the attitude state equation can be constructed by utilizing the integral relationship between angular rate and angle change. For example, a time update equation based on Kalman filtering can be established, where the state variables are roll angle and pitch angle, and the control input is the angular rate collected by the gyroscope. The change in angle can be calculated by multiplying the angular rate by the sampling time interval.

[0082] Step 420: Input the current angular rate and the attitude estimate from the previous preset time into the attitude state equation to obtain the attitude prediction value for the current time output by the attitude state equation.

[0083] Specifically, the optimal attitude estimate from the previous moment is read, combined with the angular rate data output by the gyroscope at the current moment, and substituted into the attitude state equation for calculation, thereby obtaining a current attitude prediction value based solely on the laws of physical motion. This attitude prediction value does not yet include the observation information from the accelerometer at the current moment.

[0084] Step 430: Calculate the target correction factor, which represents the confidence weight relationship between the current attitude prediction value and the current fuselage attitude data.

[0085] The confidence weight relationship refers to the degree to which the predicted value derived from the attitude state equation or the measured value directly collected by the sensor is more favored in the final result. For example, the confidence weight relationship can be the Kalman gain, which reflects the proportional relationship between the prediction error covariance and the observation noise covariance.

[0086] Specifically, calculating the target correction factor requires combining the system's prediction error covariance matrix and measurement noise covariance matrix. For example, by calculating the Kalman gain matrix, when the uncertainty in the attitude state equation is large, the correction factor will increase to make more use of the measured data; conversely, when the sensor noise is large, the correction factor will decrease to rely more on the predicted value of the attitude state equation.

[0087] Step 440: Correct the attitude prediction value and fuselage attitude data using the target correction factor to obtain smooth attitude data.

[0088] Specifically, the calculated Kalman gain is used as a weighting coefficient to calculate the residual between the attitude prediction value and the actual attitude data measured by the accelerometer at the current moment (such as the tilt angle calculated by the gravity component). The weighted residual is then used to compensate for the attitude prediction value, thereby obtaining the optimal estimate that combines the advantages of kinematic deduction and actual measurement data, i.e., smoothed attitude data.

[0089] The harvester cleaning system provided by this invention constructs an attitude state equation based on angular rate and combines it with the observation data at the current moment for fusion estimation. By utilizing the confidence degree between the predicted value of the target correction factor dynamic trade-off model and the measured value of the sensor, it can effectively suppress the error accumulation and noise interference caused by a single sensor, thereby obtaining smooth attitude data with both fast dynamic response and high static accuracy.

[0090] In another embodiment of the present invention, generating control commands based on attitude trend prediction results includes: obtaining a cleaning performance table, which includes the correspondence between attitude angle ranges and lateral adjustment amounts; if the attitude trend prediction result exceeds a preset adjustment dead zone range, then finding the target lateral adjustment amount corresponding to the attitude trend prediction result according to the cleaning performance table, and generating control commands based on the target lateral adjustment amount.

[0091] The cleaning performance table refers to a data mapping table stored in the internal memory of the control module, used to define the compensation actions that the cleaning screen should take under different inclination degrees. For example, the cleaning performance table is shown in Table 1: Table 1: Relationship between the lateral tilt angle of the harvester's cleaning system and the extension amount of the electric push rod Lateral tilt angle Overhang (60mm*percentage) Less than -12 degrees +100% [-12,-6) +70% [-6,-4) +50% [-4,-2) +20% [-2,0) 0 [0,2) 0 [2,4) -20% [4,6) -50% [6,12] -70% Greater than 12 degrees -100% The preset adjustment dead zone range refers to the angle range that is not adjusted in order to prevent the system from frequent operation. For example, the preset adjustment dead zone range can be the range of horizontal tilt angle between -2 degrees and 2 degrees.

[0092] Considering that harvesters inevitably experience slight shaking during operation, adjusting even minute angle changes would lead to frequent start-stop cycles of the actuators, shortening equipment lifespan and hindering cleaning efficiency. Therefore, this invention incorporates an adjustment dead zone, triggering adjustment only when the predicted tilt angle exceeds this range. This avoids ineffective adjustments, extends the lifespan of the actuators, and ensures the stability of the control system.

[0093] The target lateral adjustment amount refers to the specific distance the cleaning screen needs to move, which is obtained by looking up the cleaning performance table. For example, the target lateral adjustment amount can be the product of the maximum stroke of the electric push rod (e.g., 60 mm) and the percentage coefficient obtained from the table. When the tilt angle is 5 degrees, the coefficient obtained from the table is -50%, so the target lateral adjustment amount is -30 mm.

[0094] Generating control commands based on the target lateral adjustment amount refers to converting the calculated distance value into an electrical signal that drives the execution module. For example, the control module sends a data frame containing a "-30mm" position command to the electric linear actuator controller via the CAN bus, instructing the electric linear actuator to retract inward by 30 mm.

[0095] The harvester cleaning system provided by this invention, by pre-setting a cleaning performance table containing the correspondence between posture angle range and lateral adjustment amount, and setting an adjustment dead zone range, generates control commands based on the table lookup results only when the predicted posture change exceeds a certain threshold. This effectively avoids frequent malfunctions of the actuator caused by minor vibrations or slight fluctuations, reduces wear and energy consumption of the mechanical system, and ensures accurate grading compensation of the cleaning screen under significantly tilted conditions, thus achieving a balance between system stability and adjustment effectiveness.

[0096] In another embodiment of the present invention, the cleaning system of the harvester further includes a measurement feedback module; the measurement feedback module acquires the actual displacement of the cleaning screen and sends the actual displacement to the control module; the control module corrects the control command based on the actual displacement.

[0097] The measurement feedback module refers to a sensor device used to detect the amplitude of the actuator's movement. For example, the measurement feedback module can be a laser rangefinder sensor installed on the side wall of the cleaning screen or near the electric push rod, or it can be a position feedback potentiometer or Hall sensor built into the electric push rod.

[0098] The actual displacement of the cleaning screen refers to the actual distance the cleaning screen moves in the lateral direction relative to the machine frame. For example, the actual displacement of the cleaning screen can be the lateral extension data of the screen surface obtained in real time by a laser rangefinder.

[0099] The actual displacement can be transmitted to the control module via analog signals or digital communication interfaces. For example, a laser rangefinder can convert the measured distance value into a voltage signal or send it to the control module via a serial communication protocol.

[0100] Correcting control commands based on actual displacement refers to the process of building a closed-loop control system. For example, the control module calculates the difference between the target lateral adjustment amount and the current actual displacement, and adjusts the drive signal output to the electric actuator according to the difference until the actual displacement matches the target value.

[0101] Considering that the cleaning screen is easily affected by factors such as mechanical clearance, load changes, or push rod aging, which may lead to incomplete execution, this invention introduces a measurement feedback module to construct a closed-loop control circuit. This module monitors the actual position of the cleaning screen in real time and corrects for deviations, thereby eliminating execution errors, ensuring that the cleaning screen moves accurately to the predetermined compensation position, and guaranteeing the stability of the cleaning effect.

[0102] The harvester cleaning system provided by this invention acquires the actual displacement of the cleaning screen in real time through a measurement feedback module and feeds it back to the control module, thus constructing a position closed-loop control loop. This loop can monitor and correct execution errors caused by mechanical clearance, load changes, or nonlinear characteristics of the actuator in real time, thereby ensuring that the cleaning screen can move accurately to the target position. This significantly improves the control accuracy and reliability of the system and guarantees the stability of cleaning performance in complex operating environments.

[0103] In another embodiment of the present invention, the execution module includes an electric push rod, one end of which is connected to the frame of the harvester, and the other end of which is connected to a cleaning screen.

[0104] The execution module may include a lateral adjustment assembly and a cleaning screen body. In addition to the electric push rod, the lateral adjustment assembly may also include auxiliary support structures such as a fifth support arm.

[0105] An electric linear actuator refers to an electric drive device that converts the rotational motion of an electric motor into the linear reciprocating motion of a linear actuator. For example, an electric linear actuator can be a DC electric linear actuator with built-in position feedback function, and its stroke range can meet the adjustment requirements of ±60 mm.

[0106] Specifically, the electric push rod is installed in a rigid connection manner. For example, the fixed end of the electric push rod is installed on the frame of the harvester chassis, and the telescopic rod end is hinged to the fifth arm of the cleaning screen. When the push rod extends or retracts, it directly pushes the fifth arm to drive the cleaning screen to slide along the transverse guide rail.

[0107] Considering that traditional hydraulic cylinder drive solutions are low in cost but have low control precision, and require additional hydraulic pipelines and displacement sensors, the system is complex and susceptible to oil temperature, this invention uses an electric actuator as the actuating element. It utilizes the characteristics of high control precision, built-in position feedback, fast response speed and easy placement to achieve micron-level (e.g., 0.1 mm) precise adjustment of the cleaning screen position, while simplifying the system structure and reducing maintenance costs.

[0108] It should be noted that this invention has significant technical advantages in terms of hardware selection.

[0109] First, regarding the execution module, although a hydraulic cylinder can be used to drive the fifth arm, this embodiment preferably uses an electric actuator. This is because while hydraulic cylinders are less expensive, their control precision is relatively low, and they typically require an additional extension measurement sensor. The electric actuator used in this embodiment, however, has built-in extension position feedback, offering higher control precision and faster response, making it more suitable for high-frequency dynamic adjustments.

[0110] Secondly, regarding the attitude detection module, although ordinary angle sensors (based on the principle of pendulum gravity) can be used, they are easily affected by external environmental factors such as wind and vibration, resulting in inaccurate output angles. This embodiment preferably uses a MEMS inertial measurement unit that integrates an IMU and a triaxial accelerometer, which has high output accuracy, strong anti-interference ability, is not affected by external wind, and can provide high-quality raw data for trend prediction algorithms.

[0111] The harvester cleaning system provided by this invention uses an electric push rod as an actuator to directly drive the cleaning screen to move laterally relative to the frame. By utilizing the characteristics of electric push rods, such as fast response speed, high control accuracy and easy integration of position feedback, the system achieves precise and rapid adjustment of the cleaning screen position, while simplifying the mechanical structure and reducing system maintenance costs.

[0112] Figure 5 This is a flowchart illustrating the cleaning method for the harvester provided by the present invention, as shown below. Figure 5 As shown, as another optional embodiment provided by the present invention, the cleaning method of the harvester includes, but is not limited to, the following steps: Step 510: Obtain the harvester's body attitude data.

[0113] Step 520: Based on the fuselage attitude data, perform attitude trend prediction to obtain attitude trend prediction results, and generate control commands based on the attitude trend prediction results.

[0114] Step 530: Send the control command to the execution module of the harvester.

[0115] It should be noted that the harvester cleaning method provided by the present invention can execute the harvester cleaning system described in any of the above embodiments during actual operation, and this embodiment will not elaborate on this.

[0116] The harvester cleaning method provided by this invention predicts the trend of the harvester's body posture data in advance, anticipates the upcoming fluctuations of the machine body, and drives the cleaning screen to actively move. This effectively solves the problem of lateral displacement and accumulation of material on the screen surface caused by response lag, maintains a stable material flow distribution, and improves the utilization rate of the screen surface and the cleaning efficiency.

[0117] Figure 6 This is the second schematic diagram of the architecture of the cleaning system for the harvester provided by the present invention, as shown below. Figure 6 As shown, the electronic control system components of the harvester's cleaning system are connected with the control module as the core. The control module communicates with external hardware via CAN bus and processes internal logic states through internal program signal lines.

[0118] Specifically, at the input end, a three-axis gyroscope sensor communicates with the control module, sending filtered longitudinal and lateral tilt angle signals of the vehicle body to the control module in real time; a laser rangefinder sensor measures the lateral extension of the screen surface and feeds the measurement results back to the control module to achieve closed-loop control; a human-machine interface is connected to the control module, allowing the driver to manually select the control mode (such as normal uniform distribution control or intelligent uniform distribution control) and display the on and off status of the cleaning function; a vehicle speed module provides the control module with the current vehicle speed information; and an electric push rod status feedback module provides the control module with real-time feedback on the current status of the electric push rod, including movement, stop, position, or fault status.

[0119] In terms of logic control, the control module internally receives an enable status signal, which indicates whether a function is on or off. When a component malfunctions, the enable signal is off. At the output end, the control module outputs the extension amount of the electric push rod based on the processing results, i.e., the absolute extension length of the electric push rod calculated by the system. Simultaneously, it outputs a parameter adjustment enable signal, such as a rising edge signal, to notify other subsystems to modify the current value and supports self-reset. Furthermore, the control module also outputs current strategy and status information, including the current strategy mode (e.g., manual or automatic mode) and the system's operating status (e.g., ready, working, faulty, or off), thereby achieving intelligent monitoring and precise control of the entire cleaning system.

[0120] As an optional embodiment, to facilitate driver operation and monitoring, the system is also equipped with a Human Machine Interface (HMI). The driver can manually select the control mode through the HMI, such as switching between manual and automatic modes. The HMI interface displays the real-time activation and operational status of the cleaning function. For example, system status can be displayed using different colored icons: green indicates the system is ready for operation; blue indicates the system is working normally; red indicates a malfunction and functional failure; gray indicates the system is off or does not support the current crop type. Furthermore, the HMI can also display the cleaning status, achieving integrated information display.

[0121] Figure 7 This is a schematic diagram of the installation location of the detection module provided by the present invention, as shown below. Figure 7 As shown, to ensure the accuracy and real-time nature of the detection data, the three-axis gyroscope sensor 10, which serves as the attitude detection module, is installed at the center of the harvester's frame, specifically on the main frame below the harvester's cab. This installation location minimizes the interference of local vibrations on the sensor measurements and accurately reflects the overall machine's attitude changes.

[0122] The three-axis gyroscope sensor 10 is rigidly fixed to the mounting bracket by bolts and connected to the vehicle's electrical system via a wiring harness. Figure 7 As shown, the wiring harness is arranged along the frame and secured with multiple cable ties and clips to prevent wear or loosening of the wiring due to vibration during operation. The three-axis gyroscope sensor 10 is powered by 24V DC and outputs a CAN signal. It communicates with the vehicle control unit (VCU) via a standard bus interface to ensure that attitude data is transmitted stably and quickly to the control module.

[0123] Figure 8 This is the third schematic diagram of the architecture of the cleaning system for the harvester provided by the present invention, as shown below.Figure 8 The diagram illustrates the mechanical structure and installation relationship of the core actuator of the cleaning system. The entire cleaning system mainly includes an upper screen box component 20, a drive component 30, and a transverse vibration component 40. The upper screen box component 20 consists of a multi-layer screen structure and has interfaces on its side walls for installation and adjustment. The drive component 30 is located at the end of the cleaning system, providing basic operating power for the cleaning operation. The transverse vibration component 40, as the core of the transverse adjustment assembly, mainly consists of an electric push rod and a fifth support arm. The electric push rod of the transverse vibration component 40 is located below the upper screen box component 20, with one end hinged to the harvester's chassis frame via a connector, and the other end connected to the fifth support arm. The fifth support arm, as an intermediate transmission rod, spans the bottom of the cleaning screen and is rigidly connected to the upper screen box component 20.

[0124] When the control system issues a command, the electric push rod in the lateral vibration component 40 extends or retracts, driving the fifth support arm to move the entire upper screen box component 20 laterally perpendicular to the harvester's forward direction. This structural design allows the upper screen box component 20 to achieve dynamic displacement compensation in both directions according to the machine's tilt, for example, adjusting within a range of ±60 mm, thereby counteracting the gravitational component caused by the side slope and ensuring that the material remains evenly distributed on the screen surface. Simultaneously, the side wall of the upper screen box component 20 also has reserved positions for installing laser rangefinder sensors to monitor the lateral extension in real time and provide feedback to the control system.

[0125] Figure 9 This is the fourth schematic diagram of the architecture of the cleaning system for the harvester provided by the present invention, as shown below. Figure 9 As shown, this illustrates the complete workflow and hardware correspondence of the cleaning system, from data acquisition and algorithm processing to execution control and feedback closed loop.

[0126] First, the attitude detection module (such as an IMU) collects the harvester's angular velocity and tilt angle data in real time and transmits this raw data to the trend prediction module. The trend prediction module processes the data using built-in algorithms such as Kalman filtering to predict future tilt angle trends.

[0127] Subsequently, the predicted tilt angle is sent to the control module (i.e., the VCU controller). The control module stores cleaning performance tables for different crops, such as cleaning performance tables for wheat, corn, and other crops. The control module looks up and matches the tables according to the crop type of the current operation to calculate the corresponding target control parameters.

[0128] Based on this parameter, the control module sends a command to the actuator regarding the extension amount of the electric push rod. The actuator includes an electric push rod and a fifth arm mounted on the harvester chassis. The electric push rod drives the fifth arm according to the command, thereby causing the cleaning screens (including the upper screen and the pre-screen) to move in a direction perpendicular to the harvester (lateral).

[0129] To ensure adjustment accuracy, a laser rangefinder is installed on the side of the cleaning screen. This laser rangefinder measures the lateral displacement of the cleaning screen relative to the harvester chassis in real time and feeds the measurement results back to the control module as a laser rangefinder closed-loop signal, thus forming a high-precision position closed-loop control system to ensure that the screen surface can be accurately and dynamically compensated according to the predicted tilt trend.

[0130] As an optional embodiment, in terms of mechanical connection, the electric push rod is connected to the fifth arm. After receiving the control signal from the VCU main control module, the electric push rod adjusts its extension distance accordingly. The electric push rod drives the fifth arm, which in turn drives the cleaning screen body. A laser rangefinder sensor installed on the side wall of the screen surface measures the lateral extension of the screen surface and feeds it back to the VCU controller for closed-loop adjustment. The VCU main control module uses a high-performance computing unit, with a preset adjustment range of ±12 degrees for the lateral angle. When the function is activated, lateral adjustment is initiated immediately when the screen is within the adjustment range.

[0131] Figure 10 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 10 As shown, the electronic device may include a processor 1010, a communications interface 1020, a memory 1030, and a communication bus 1040. The processor 1010, communications interface 1020, and memory 1030 communicate with each other via the communication bus 1040. The processor 1010 can call logical instructions from the memory 1030 to execute the harvester's cleaning system. This method includes: acquiring the harvester's body attitude data; performing attitude trend prediction based on the body attitude data to obtain the attitude trend prediction result; generating control instructions based on the attitude trend prediction result; and sending the control instructions to the harvester's execution module.

[0132] Furthermore, the logical instructions in the aforementioned memory 1030 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0133] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the harvester cleaning system provided by the above methods. The method includes: acquiring the harvester's body posture data; performing posture trend prediction based on the body posture data to obtain the posture trend prediction result; generating control instructions based on the posture trend prediction result; and sending the control instructions to the harvester's execution module.

[0134] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements a cleaning system for a harvester provided by the methods described above. The method includes: acquiring the harvester's body posture data; performing posture trend prediction based on the body posture data to obtain a posture trend prediction result; generating control instructions based on the posture trend prediction result; and sending the control instructions to the harvester's execution module.

[0135] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0136] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cleaning system for a harvester, characterized in that, It includes an attitude detection module, a control module, and an execution module; The attitude detection module acquires the harvester's body attitude data and sends the body attitude data to the control module; The control module performs attitude trend prediction based on the fuselage attitude data, obtains attitude trend prediction results, generates control commands based on the attitude trend prediction results, and sends the control commands to the execution module. The execution module drives the cleaning screen of the harvester to move based on the control commands.

2. The cleaning system for a harvester according to claim 1, characterized in that, The attitude trend prediction based on the fuselage attitude data, to obtain the attitude trend prediction result, includes: The fuselage attitude data is smoothed to obtain smoothed attitude data; The smoothed attitude data is input into the attitude trend prediction model to obtain the attitude trend prediction result output by the attitude trend prediction model; The posture trend prediction model is trained based on historical posture sequence samples.

3. The cleaning system for a harvester according to claim 2, characterized in that, The posture trend prediction model is obtained based on the following training steps: Collect historical attitude sequence samples, which include the harvester's roll angle sequence, pitch angle sequence, angular rate sequence, and acceleration data during historical operating periods; Determine the future attitude label corresponding to the historical attitude sequence sample, wherein the future attitude label is the actual tilt angle of the harvester at a preset time interval after the historical attitude sequence sample collection time; The training process is executed iteratively until the preset termination condition is met. The training process includes: Input the historical attitude sequence samples into the attitude trend prediction model to obtain the predicted tilt angle output by the attitude trend prediction model; Calculate the loss between the predicted tilt angle and the future attitude label; The parameters of the attitude trend prediction model are updated based on the loss.

4. The cleaning system for a harvester according to claim 2, characterized in that, The smoothing process of the fuselage attitude data to obtain smoothed attitude data includes: The attitude state equation of the harvester is constructed, and the attitude state equation is established based on the angular rate in the machine attitude data; The current angular rate and the attitude estimate from the previous preset time are input into the attitude state equation to obtain the attitude prediction value at the current time output by the attitude state equation. Calculate the target correction factor, which represents the confidence weight relationship between the current attitude prediction value and the current fuselage attitude data; The target correction factor is used to correct the attitude prediction value and the fuselage attitude data to obtain the smoothed attitude data.

5. The cleaning system for a harvester according to claim 1, characterized in that, Based on the attitude trend prediction results, control commands are generated, including: Obtain a cleaning performance table, which includes the correspondence between attitude angle ranges and lateral adjustment amounts; If the attitude trend prediction result exceeds the preset adjustment dead zone range, then the target lateral adjustment amount corresponding to the attitude trend prediction result is found according to the cleaning performance table, and the control command is generated according to the target lateral adjustment amount.

6. The cleaning system for a harvester according to claim 1, characterized in that, The cleaning system of the harvester also includes a measurement feedback module; The measurement feedback module obtains the actual displacement of the cleaning screen and sends the actual displacement to the control module; The control module modifies the control command based on the actual displacement.

7. The cleaning system for a harvester according to claim 1, characterized in that, The execution module includes an electric push rod, one end of which is connected to the frame of the harvester, and the other end of which is connected to the cleaning screen.

8. A cleaning method for a harvester, characterized in that, include: Acquire the harvester's body attitude data; Based on the fuselage attitude data, attitude trend prediction is performed to obtain attitude trend prediction results, and control commands are generated based on the attitude trend prediction results. The control commands are sent to the execution module of the harvester.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the cleaning method of the harvester as described in claim 8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the cleaning method of the harvester as described in claim 8.

Images