Method and device for controlling a bulldozer blade

Abstract

The embodiment of the application discloses a bulldozer blade control method and system. The method comprises the following steps: predicting a change trend of a vehicle body pitch angle of the bulldozer based on attitude sensing data; when a positive mutation of the vehicle body pitch angle exceeding a first threshold is predicted according to the change trend of the vehicle body pitch angle, active compliance control is performed; wherein the positive mutation of the vehicle body pitch angle is a change of a vehicle head sinking relative to the vehicle body of the bulldozer within a preset time slice; the active compliance control comprises: sending a blade lifting instruction to a lifting cylinder of the blade to make the blade lift upward to release elastic potential energy accumulated on a blade edge of the blade; in a process of the vehicle body pitch angle falling back under the action of gravity, the blade is controlled by using falling back inertia and a blade falling instruction to make the blade re-contact a soil layer with a preset cutting force.

Description

Technical Field

[0001] This invention relates to the field of construction technology, and in particular to a method and device for controlling the blade of a bulldozer. Background Technology

[0002] Bulldozers are core equipment in earthmoving projects, and the accuracy of their blade control directly affects construction quality and efficiency. Current automatic control technology for bulldozer blades generally adopts a position tracking control paradigm: the system compares the difference between the current blade elevation and the design elevation in real time, and uses a proportional-integral-derivative (PID) controller to drive the hydraulic system to correct the blade's posture, causing the blade to move along a preset trajectory. This scheme achieves good control results under ideal rigid ground conditions. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method and apparatus for controlling a bulldozer blade. The technical solution of the present invention is implemented as follows: A first aspect provides a method for controlling a bulldozer blade, characterized by comprising: predicting the trend of the bulldozer's body pitch angle based on attitude sensing data; when a positive abrupt change in the body pitch angle is predicted to exceed a first threshold based on the trend of the body pitch angle change, executing active compliance control; wherein the positive abrupt change in the body pitch angle is the change in the downward movement of the bulldozer's front end relative to the bulldozer's body within a preset time slice; the active compliance control comprises: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, thereby releasing the elastic potential energy accumulated on the blade's cutting edge; during the process of the body pitch angle falling back under gravity, controlling the blade using the falling inertia and the blade falling command, so that the blade re-contacts the soil layer with a preset cutting force.

[0004] The second aspect provides a bulldozer blade control device, comprising: a prediction module for predicting the trend of the bulldozer's body pitch angle based on attitude sensing data; and an execution module for executing active compliance control when a positive abrupt change in the body pitch angle is predicted to exceed a first threshold based on the trend of the body pitch angle change; wherein the positive abrupt change in the body pitch angle is the change in the downward movement of the bulldozer's front end relative to the bulldozer's body within a preset time slice; the active compliance control includes: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, thereby releasing the elastic potential energy accumulated on the blade's cutting edge; and during the process of the body pitch angle falling back under gravity, controlling the blade using the falling inertia and the blade falling command to make the blade re-contact the soil layer with a preset cutting force.

[0005] The third aspect provides a computer-readable storage medium storing computer-executable instructions; after being executed by a processor, the computer-executable instructions are able to implement the method provided by any of the technical solutions of the first aspect.

[0006] This disclosed embodiment no longer uses the adversarial control method based on detecting deviations and reverse corrections in the prior art, but instead adopts active compliant control based on predicting trends and conforming to the same direction. When a sudden change in the vehicle's pitch angle is predicted, a lifting command is actively sent to the lifting cylinder, causing the blade to rise and releasing the elastic potential energy accumulated on the blade edge, thereby eliminating the source of torque driving the vehicle upward. After the vehicle naturally falls back under the influence of gravity, the blade is commanded to re-contact the soil layer with a preset cutting force, utilizing the vehicle's downward inertia. In this way, the positive feedback loop is fundamentally cut off, reducing the attitude overshoot under hard soil cutting conditions and shortening the convergence time of the vehicle's pitch angle from the disturbance peak to the stable range.

[0007] This embodiment of the invention can predict the trend of vehicle pitch angle change based on attitude sensing data, and trigger active compliance control before the change of attitude occurs, realizing the transformation from passive correction to active prediction, and improving the control system's ability to cope with sudden operating conditions.

[0008] By actively lifting the blade to release elastic potential energy, hard impact between the blade edge and hard soil layers is avoided, reducing the impact load on the blade edge. Simultaneously, it avoids high power output from the hydraulic system under pressure buildup, reducing fatigue damage to the hydraulic system. Calculations show that this embodiment can extend the blade edge replacement cycle and reduce the hydraulic system failure rate.

[0009] The embodiments disclosed herein are based entirely on the existing sensor configuration of bulldozers (inertial measurement unit, attitude sensor), without the need to add any additional hardware devices. The above performance improvement can be achieved simply by upgrading the algorithm, which has good economic benefits and promotional value. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a bulldozer blade control method provided in an embodiment of the present invention; Figure 2 A flowchart illustrating a bulldozer blade control method provided in an embodiment of the present invention; Figure 3A A schematic diagram of a bulldozer provided for an embodiment of the present invention; Figure 3BA schematic diagram of a bulldozer provided for an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a bulldozer blade control device provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0011] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0012] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used 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 described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0013] When bulldozers operate under complex conditions, the interaction force between the blade and the soil exhibits strong nonlinearity, time-varying characteristics, and abrupt changes. Control methods in related technologies primarily rely on real-time feedback of the current state, i.e., detecting and correcting deviations. This approach reveals significant shortcomings in the following scenarios: First, when the blade suddenly cuts into hard soil or collides with underground obstacles, the elastic potential energy is released in a very short time, causing the vehicle to buckle or the blade to bounce, and feedback control cannot respond in time; second, different soil types result in vastly different optimal control parameters for the blade, and a single parameter cannot accommodate all working conditions; third, in precision leveling operations, traditional proportional-integral-derivative (PID) control struggles to simultaneously meet the contradictory requirements of rapid approach and avoiding overcutting.

[0014] To address the aforementioned technical problems, this disclosure proposes a predictive, identification, adaptation, and compliant intelligent control architecture applicable to automatic or semi-automatic bulldozers, such as unmanned bulldozers. In this disclosure, firstly, by predicting the abrupt changes in vehicle posture, energy is actively released before impact, converting destructive impact into usable inertial soil-penetrating power. Secondly, by identifying soil type and its layered structure online, nonlinear switching and smooth transition of control parameters are driven. Thirdly, a control model of a virtual elastic layer is constructed, and a stiffness sign switching mechanism is introduced to achieve free conversion between compliant approach and active pull-in during the leveling stage. Finally, flutter signals are used to reduce local resistance, allowing the blade to passively level under gravity, overturning the energy consumption logic of traditional active leveling. The above inventive concept is elaborated in detail below through several embodiments.

[0015] To better illustrate the embodiments of this disclosure, the relevant terms will be introduced below: Active adaptive control: A sequence of control actions actively executed before the impact occurs, including two stages: raising the shovel to release elastic potential energy and using the vehicle's inertia to re-enter the soil. For example, if it is predicted that the vehicle's front end will sink, the shovel is raised 3 centimeters 0.1 seconds in advance, and then lowered again with a smaller force after the vehicle's front end has fallen back down.

[0016] Elastic potential energy: The mechanical energy stored when the blade of a shovel comes into contact with the soil due to the elastic deformation of both sides. For example, when a shovel pushes against a rock, causing the connecting rod to bend by 0.5 mm, the energy stored in this bending is elastic potential energy. When the rock breaks, this energy is released, causing the shovel to bounce.

[0017] Equivalent cutting force: The combined force acting on the blade edge, obtained by converting the pressure of each cylinder through a dynamic model. For example, the soil resistance experienced by the blade is determined based on the pressure values ​​of the left and right tilting cylinders and the lifting cylinder, combined with the force transmission coefficient of the linkage mechanism.

[0018] Mixed soil: A soil structure containing at least two soil layers with different mechanical properties within the cutting depth range of the shovel. For example, a top 20 cm layer of loose backfill soil and a bottom 30 cm layer of dense silty clay constitute mixed soil.

[0019] Step increase: A significant increase in equivalent cutting force that occurs and remains at a high level over a short period of time, as opposed to a momentary spike. For example, the average change in equivalent cutting force is greater than the average threshold. For instance, the cutting force jumps from 100 kN to 150 kN in 0.05 seconds and stabilizes at that level, rather than instantaneously surging to 300 kN and then immediately falling back.

[0020] Virtual elastic layer: A virtual force field region constructed in the controller algorithm, located above the target design elevation. The control behavior after the blade enters is dominated by this force field. For example, by setting a virtual elastic layer within 5 cm above the target elevation, the closer the blade gets to the target surface, the greater the reverse thrust it experiences, achieving a compliant approach.

[0021] Positive stiffness mode: A working mode of the virtual elastic layer. In this mode, the closer the blade gets to the target design elevation, the smaller the upward thrust output of the lifting cylinder. For example, if the target elevation is 100.00 meters, the cylinder output is 1500N upward when the blade is at 99.97 meters, and 500N upward when the blade is at 99.99 meters. The output decreases as the blade approaches, causing it to decelerate as it approaches.

[0022] Negative stiffness mode: A working mode of the virtual elastic layer. In this mode, when the blade is slightly higher than the target design elevation, the lifting cylinder outputs a downward pulling force. The magnitude of the pulling force increases with the distance between the blade and the target design elevation, actively pulling the blade towards the target surface to solve the under-digging problem. For example, if the target elevation is 100.00 meters, the cylinder outputs a downward pulling force of 800N when the blade is at 100.03 meters (under-digging 3 cm), and 200N when the blade is at 100.01 meters (under-digging 1 cm). The force increases with the distance, actively pulling the blade closer to the target surface, while over-damping control prevents oscillation.

[0023] Overdamped state: A system state with a damping ratio greater than 1, where the response has no overshoot and oscillations. For example, in negative stiffness mode, if the damping coefficient is set to a predetermined multiple of the critical damping (e.g., 1.5, 1.8, or 2), no reciprocating oscillations will occur when the blade moves to the target surface.

[0024] The first chatter signal is a high-frequency, micro-amplitude vibration superimposed on the control signal of the blade lifting cylinder when operating in hard gravel or rock strata. Its function is to disrupt the static friction between the hard material and the blade edge, reducing cutting resistance and assisting the blade in cutting. For example, when the blade cuts into gravel, the controller superimposes a 100Hz sine wave signal with an amplitude of 8% of the rated pressure onto the blade lowering command from the lifting cylinder. This high-frequency vibration transforms the static friction between the gravel and the blade edge into dynamic friction, reducing the cutting force by approximately 30%.

[0025] The second flutter signal is a vibration signal superimposed on the blade control signal during the interlayer transition of the mixed soil. Its function is to pre-treat the interface of the hard soil layer with micro-vibration, causing micro-cracks on the surface of the interface and reducing the peak impact force when the blade crosses the interface. For example, just before the blade cuts from the clay layer into the gravel layer, the controller identifies the natural frequency of the gravel layer as 32Hz and superimposes a flutter signal with a frequency of 32Hz and an amplitude of 8% of the rated pressure on the blade for 0.5 seconds, causing a softening layer of about 2 mm thick to form on the surface of the gravel interface, reducing the cross-layer impact force from 200kN to 110kN.

[0026] The difference between the second flutter signal and the first flutter signal is that the frequency of the second flutter signal is dynamically adjusted according to the natural frequency of the soil layer (rather than a fixed high frequency), and the timing of its application is precisely controlled at the moment before crossing the layer (rather than the entire operation process), with the aim of "interface softening" (rather than continuous drag reduction).

[0027] The third chatter signal: This vibration signal is applied to the tilting cylinder on the side with less force when the blade tilts laterally. Its function is to selectively reduce the soil cutting resistance on that side, allowing the blade on that side to sink naturally under gravity, achieving passive leveling. For example, if the blade tilts to the right, the force on the left side is less than on the right. The controller applies a chatter signal with a frequency of 20Hz and an amplitude of 10% of the rated pressure to the left tilting cylinder. The resistance on that side decreases by approximately 40%, and the left blade automatically sinks under gravity, restoring balance between the left and right sides within 0.3 seconds.

[0028] The difference between the third and second chatter signals: The third chatter signal is applied to the tilting cylinder (rather than the lifting cylinder) and acts on a single blade (rather than the whole blade), with the purpose of achieving gravity-assisted leveling (rather than reducing resistance or softening the interface).

[0029] Impact-type entry: This strategy involves raising the blade to a preset height and then using a combination of gravity and hydraulic thrust to accelerate its descent and impact hard rock formations. For example, after raising the blade to a height representing a certain distance (e.g., 15 cm), the lowering valve is released and a hydraulic impact is applied, causing the blade to strike the rock layer at a high speed.

[0030] In some cases, the electro-hydraulic control system of a bulldozer includes at least an attitude sensor (e.g., a six-axis inertial measurement unit, IMU) mounted on the vehicle body, pressure sensors mounted on the blade lifting cylinder and tilting cylinder, displacement sensors mounted on the blade and its linkage mechanism, and a central controller. The controller is communicatively connected to the aforementioned sensors and the electro-hydraulic proportional valves of the blade hydraulic system, and records each operation command issued by the user (i.e., the operator or the autopilot system) and the blade's response data during each operation, forming an operation log.

[0031] Please see Figure 1 , Figure 1This is a flowchart illustrating a bulldozer blade control method according to an embodiment of the present invention. The bulldozer blade control method described in this embodiment includes the following steps: 101. Predict the trend of bulldozer body pitch angle change based on attitude sensing data.

[0032] In some embodiments, the attitude sensor is a sensor used to measure the attitude angle of the bulldozer body in space, typically employing a six-axis inertial measurement unit (IMU) that integrates a three-axis accelerometer and a three-axis gyroscope. The accelerometer calculates the static pitch angle of the vehicle body by measuring the components of gravitational acceleration in each axis, while the gyroscope dynamically tracks the rate of change of the pitch angle by measuring the angular velocity of the vehicle body around each axis.

[0033] In practical applications, the controller fuses data from the accelerometer and gyroscope using data fusion algorithms (such as complementary filtering or Kalman filtering) to obtain stable and sensitive dynamic pitch angle information. Based on continuously collected pitch angle data, the controller calculates the pitch angular velocity and pitch acceleration using differential calculations, and then predicts the trend of the vehicle's pitch angle in the next preset time slice (e.g., 100ms to 250ms).

[0034] If the prediction results indicate that the front of the vehicle will sink rapidly relative to the vehicle body beyond the first threshold, the system will determine that a positive change is about to occur, thereby triggering active compliance control in advance.

[0035] In some embodiments, the attitude sensor is typically installed in the rear part of the bulldozer frame near the center of gravity. This location can more accurately reflect the overall attitude changes of the vehicle body, while avoiding excessive interference from local high-frequency vibrations at the front of the vehicle on the measurement signal.

[0036] In some embodiments, after the bulldozer is started and operations begin, the controller (e.g., an onboard controller or a remote controller) continuously receives data from attitude sensors at a frequency of at least 50 Hz. Instead of directly using the current pitch angle for deviation calculation, the controller first performs a prediction step. Specifically, the controller inputs pitch angle time-series data from the most recent time window (e.g., 0.5 seconds) into a prediction model. This model can use a sliding window linear fitting method to calculate the instantaneous rate of change of the pitch angle and its acceleration, thereby extrapolating the pitch angle change trajectory within the next 0.2 to 0.3 seconds. This prediction mechanism allows the controller to anticipate changes in the vehicle's attitude before they actually occur, fundamentally overcoming the inherent hysteresis of traditional feedback control.

[0037] 102. When the sudden change in the vehicle pitch angle is predicted to exceed the first threshold based on the trend of the vehicle pitch angle change, active compliance control is executed.

[0038] In some embodiments, the positive abrupt change in the vehicle body pitch angle is the change in the downward slope of the bulldozer's front end relative to the bulldozer's body within a preset time slice.

[0039] In some embodiments, the preset time slice is a time window for determining whether a sudden change in pitch angle occurs, and its value ranges from 100ms to 500ms, for example, 200ms or 300ms. In some embodiments, the preset time slice may be one or more detection time windows for attitude sensing data. In summary, the change in pitch angle of the vehicle front relative to the vehicle body is calculated within the time window corresponding to the preset time slice to distinguish between slow pitch during normal operation and sudden sinking under dangerous conditions.

[0040] In some embodiments, the cab is the structural area in front of the connection point between the bulldozer's blade arm and the frame, and the body is the bulldozer's main frame. The cab's relative depression to the body refers to the rapid decrease in the difference between the cab's local pitch angle and the main frame's pitch angle within a preset time period. The first threshold refers to the upper limit of the change in this difference within the preset time period. When the change exceeds this threshold, it is determined to be a positive abrupt change. The threshold ranges from 2 to 5 degrees. It is worth noting that the specific value of the first threshold may be related to the bulldozer's model and its own physical parameters (e.g., the variable angles of the cab and body, the height of the cab, etc.). In some embodiments, such a change in vehicle pitch angle may be a positive change in vehicle pitch angle exceeding a first threshold, in which active compliance control is performed.

[0041] In some embodiments, if the sudden change in vehicle pitch angle is a negative change and / or the positive change does not exceed a first threshold, then active compliance control is not performed. When specific active compliance control is not performed, conventional control can be applied to the bulldozer's blade.

[0042] In some embodiments, a negative mutation may include: the bulldozer cab rising up within a preset time slice.

[0043] In some embodiments, the controller evaluates the predicted trajectory in real time. When the prediction shows that the vehicle's pitch angle will change abruptly within a preset time slice, and the estimated magnitude of this change exceeds a first threshold (e.g., 3 to 5 degrees, the specific value of which can be calibrated according to the vehicle model and working conditions), the controller determines that a hard obstacle has been encountered in front of the blade edge or that the soil stiffness has increased abruptly, resulting in a large accumulation of elastic potential energy at the blade edge. At this point, if the controller waits for the actual impact to occur before responding, the vehicle body will have already undergone uncontrollable attitude changes, and the leveling accuracy will be severely affected. Therefore, the controller immediately triggers an active compliant control process.

[0044] In some embodiments, active compliance control includes: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, thereby releasing the elastic potential energy accumulated on the blade's cutting edge; and controlling the blade using the falling inertia and the blade's falling command during the descent of the vehicle body's pitch angle under gravity, so that the blade re-contacts the soil layer with a preset cutting force.

[0045] For example, firstly, the controller sends a lifting command to the electro-hydraulic proportional valve of the blade lifting cylinder, causing the blade to lift upwards by a preset small displacement (e.g., 3 to 5 centimeters). This lifting action is not a simple upward movement, but includes a calibrated lifting speed curve: the cylinder gap is eliminated at a relatively fast initial stage, followed by an effective lifting at a moderate speed. The physical essence of this lifting action is the active release of the elastic potential energy accumulated between the blade and the soil layer due to compression deformation. In actual tests, when the blade cuts into dense silty clay and encounters buried rocks, if the blade is not actively lifted, the elastic potential energy will be violently released the moment the rocks break, causing the blade to bounce upwards by more than 10 centimeters; however, with the active lifting of this invention, the energy is released in a controllable manner, the blade only undergoes a small displacement, and the vehicle's posture is almost unaffected.

[0046] For example, secondly, after the blade is raised, the cutting edge temporarily loses contact with the soil. The vehicle body, losing the upward support of the blade, naturally sinks under its own weight, and the pitch angle of the vehicle head gradually returns to a horizontal position. The controller monitors the pitch angle's decline in real time. When it detects that the pitch angle has recovered to a certain percentage (e.g., 90%) of its pre-abrupt state, the controller determines that the vehicle's potential energy has been fully accumulated. Subsequently, the controller uses the vehicle's downward inertia to send a blade-lowering command to the lifting cylinder. This command includes a preset cutting force parameter (a pre-set limit in the form of an upper limit of cylinder pressure). Under the combined action of the vehicle's weight and hydraulic drive force, the blade re-enters the soil at a controlled speed and force. Compared to the blind search for a new cutting point after the blade leaves the soil in traditional methods, this embodiment utilizes the same wave energy cycle to complete the entire process of release, return, and re-entry. This not only eliminates the adverse effects of impact on leveling accuracy but also converts the originally harmful elastic potential energy and vehicle weight into soil-entry power.

[0047] In one feasible embodiment, if the controller predicts a new abrupt change in pitch angle during the execution of active compliance control (e.g., a second hard obstacle below the soil layer), the controller can immediately superimpose a new lifting command before a single lifting and lowering operation is completed, forming a multi-level compliance sequence. This nested compliance control can handle extreme stepped hard soil layers, thereby improving operational efficiency and quality.

[0048] In some embodiments, the preset cutting force parameter is pre-set in the form of an upper pressure limit for the lifting cylinder. Its physical meaning is: the maximum working pressure that the control system allows the lifting cylinder to reach when the blade enters the soil for the second time. This pressure value directly determines the maximum cutting force exerted by the blade edge on the soil layer. Specifically, the calibration of this upper pressure limit is based on a combination of one or more of the following factors: Soil type and operating conditions vary significantly, with different soil types exhibiting different cutting resistance. For soft sandy soils, a lower pressure limit (e.g., 5 MPa) can be set to prevent excessive cutting by the blade, which could cause the bulldozer's front wheels to lift off the ground or the tracks to slip. For hard clay or gravel-mixed soils, a higher pressure limit (e.g., 12 MPa) is required to provide sufficient cutting power to ensure operational efficiency.

[0049] The upper limit of the pressure setting for the bulldozer's overall parameters must be matched with the bulldozer's overall weight, engine power, and rated pressure of the hydraulic system. In principle, this upper limit should not exceed 80% of the rated working pressure of the lifting cylinder to maintain a safety margin for the hydraulic system; simultaneously, it should ensure that the maximum cutting force generated is less than the maximum traction force of the bulldozer's traveling mechanism to prevent track slippage during cutting.

[0050] The blade's posture and entry angle, as well as the cutting force parameters, are coupled with the blade's tilt angle (pitch angle, tilt angle). Under the same lifting cylinder pressure, different contact areas and force directions between the blade's cutting edge and the soil layer will result in different actual cutting effects. Therefore, the pressure limit is usually stored in the controller in the form of a "pressure-tilt angle two-dimensional mapping table," and the system dynamically calls the corresponding pressure limit value based on the current actual posture of the blade.

[0051] For operations requiring high precision, such as fine leveling and slope trimming, a lower pressure limit (e.g., 3-5 MPa) should be set to allow the blade to trim the soil surface with a "light touch and slide" motion, avoiding large-scale cutting and damage to the already formed cross-section. For operations requiring high efficiency, such as coarse pushing and material accumulation, a higher pressure limit (e.g., 10-15 MPa) can be set to fully utilize the blade's cutting ability.

[0052] At the control execution level, this pressure upper limit serves as a limiting signal for the proportional relief valve of the lifting cylinder or the pressure closed-loop control module. When the controller issues a command to lower the blade, the hydraulic system supplies oil to the large chamber of the lifting cylinder, while simultaneously monitoring the feedback value from the cylinder pressure sensor in real time. If the actual pressure reaches the preset upper limit, the oil supply flow is automatically limited or the relief valve is opened to maintain the pressure within the limit, thereby ensuring that the blade cuts into the soil with the preset maximum cutting force. This avoids secondary impacts caused by excessive cutting and also prevents work interruptions due to insufficient cutting power.

[0053] Figures 3A to 3BThe bulldozer shown is the one corresponding to the bulldozer blade control method of this disclosure embodiment. Compared with the method in related technologies that relies on the driver's experience to manually control the blade's entry force, the bulldozer blade control method provided in this disclosure embodiment achieves quantitative control and automatic amplitude limiting of the secondary entry process through preset entry force parameters, significantly reducing the difficulty of operation and operational inconsistencies.

[0054] Please see Figure 2 , Figure 2 This is a flowchart illustrating another bulldozer blade control method provided in an embodiment of the present invention. Before the active conformal control is executed, the method further includes the steps of online soil type identification and parameter adaptation.

[0055] 201. Collect pressure data of the hydraulic system of the shovel.

[0056] During the operation of the shovel, the controller synchronously collects two signals at a sampling rate of no less than a specific frequency (e.g., 100 Hz): one is the vehicle attitude data from the attitude sensor, and the other is the pressure data from the shovel hydraulic system.

[0057] In some embodiments, the bulldozer's lifting cylinder is a double-acting, single-piston-rod hydraulic cylinder. The piston inside the cylinder divides it into two chambers: the chamber through which no piston rod passes is called the large chamber (rodless chamber), which has a larger effective working area; the chamber through which the piston rod passes is called the small chamber (rod chamber), which has a smaller effective working area. When hydraulic oil enters the large chamber, the cylinder extends, and the blade is lifted upwards; when hydraulic oil enters the small chamber, the cylinder contracts, and the blade retracts downwards. The pressure in the large chamber reflects the force used by the lifting cylinder to push the blade upwards, while the pressure in the small chamber reflects the force used to prevent the blade from falling or to actively press down on the blade. These two pressures are typically detected in real-time by pressure sensors installed at the cylinder's inlet and outlet. The sensors convert the pressure signals into electrical signals, which are then sent to the controller to determine the current force state of the blade and make control decisions such as lifting, hovering, or actively lowering. In addition, the bulldozer also has two tilting cylinders, each with its own large and small chamber pressures, used to control the blade's tilt angle and monitor eccentric loading. At this time, the pressure data to be detected may include, but is not limited to: the pressure in the large chamber of the lifting cylinder, the pressure in the small chamber, and the pressure in the two tilting cylinders on the left and right.

[0058] 202. Calculate the equivalent cutting force on the shovel based on the pressure data, and extract the feature vector based on the time series data of the equivalent cutting force.

[0059] In some embodiments, due to the different force-bearing areas of different cylinders and the amplification or reduction effect of force in the linkage mechanism, a simple pressure value cannot directly characterize the interaction intensity between the blade and the soil. Therefore, the controller first performs a force calculation step: based on the dynamic model of the bulldozer's working device, it converts the pressure of each cylinder into the equivalent cutting force at the blade edge. Specifically, the controller calculates the cutting force on the left and right sides separately, and the sum of the two is the total equivalent cutting force. This calculation process also considers the current height and tilt angle of the blade, because the force transmission efficiency of the cylinders differs under different postures.

[0060] After obtaining the time-series waveform of the equivalent cutting force, the controller further extracts a feature vector that characterizes the soil's mechanical properties. Within a sliding time window, the controller calculates multiple statistics and signal characteristics: the average value of the equivalent cutting force within the window, reflecting the average hardness of the current soil layer; variance or standard deviation, reflecting the homogeneity of the soil layer, with a larger variance indicating the presence of discrete hard particles in the soil layer; peak frequency, obtained through spectral analysis of the waveform, showing significant differences in the frequency of cutting force fluctuations among different soil types, with loose soils exhibiting a relatively flat cutting force spectrum, while dense silty clay shows energy concentration at specific frequencies; and impact energy, i.e., the integral of the waveform envelope over time, which is particularly sensitive to sudden impacts from hard objects. These features collectively constitute a multidimensional feature vector.

[0061] 203. Input the feature vector into the classifier to identify the soil type for the current operation.

[0062] The aforementioned feature vectors are input into a pre-trained classifier. This classifier can be implemented using a support vector machine, random forest, or lightweight neural network, and is trained using a large amount of field-collected calibration data before leaving the factory. The classifier's output labels include four categories: loose backfill soil, dense silty clay, hard gravelly soil or rock, and mixed soil.

[0063] Mixed soil refers to soil layers containing at least two distinctly different mechanical properties within the planned cutting depth of the shovel. For example, the upper layer might be 30 cm of loose backfill soil, and the lower layer 20 cm of hard gravelly soil; or the upper layer might be topsoil with humus, the middle layer dense silty clay, and the bottom layer weathered rock. When the classifier detects frequent jumps in feature vectors between class boundaries or consistently low confidence levels, it classifies the soil as mixed soil.

[0064] 204. Adjust the stiffness coefficient and damping coefficient of the control blade according to the current soil type.

[0065] Based on the identified soil type, the controller adjusts the stiffness and damping coefficients used to control the blade. The stiffness coefficient primarily affects the blade's response to positional deviations, while the damping coefficient affects the system's ability to suppress sustained resistance.

[0066] Specifically, for loose backfill, the controller uses a first stiffness coefficient and a first damping parameter to control the blade. The resistance of loose backfill increases slowly with depth and is prone to lateral flow. Excessive stiffness can cause the blade to become oversensitive, amplifying minor terrain undulations into violent hydraulic cylinder movements; excessive damping, on the other hand, can hinder the blade from following terrain changes, resulting in missed areas. A low-stiffness, low-damping configuration allows the blade to float on the soil surface, relying primarily on its own weight to complete the cutting.

[0067] For dense silty clay, the controller employs a second stiffness coefficient and a second damping parameter to control the blade, where the second stiffness coefficient is higher than the first stiffness coefficient, and the second damping parameter is higher than the first damping parameter. Dense silty clay is characterized by its adhesiveness and a certain degree of plasticity. After the blade cuts in, it generates continuous resistance and easily forms clods of soil at the cutting edge. The medium-to-high stiffness ensures that the blade does not passively retreat when facing continuous resistance, maintaining the set cutting depth; the medium-to-high damping suppresses high-frequency blade chatter caused by intermittent soil clod shedding.

[0068] For hard gravel or rock strata, the controller uses a third stiffness coefficient and a third damping parameter, superimposed with a first chatter signal to control the blade. The third stiffness coefficient is higher than the second stiffness coefficient, the third damping parameter is higher than the second damping parameter, and the frequency of the first chatter signal is higher than the preset frequency. High stiffness and high damping are necessary conditions to ensure that the blade can maintain positional stability under point impacts. A system lacking sufficient stiffness will rebound with each impact on hard rocks, leading to a sharp drop in work efficiency. The superimposed first chatter signal has high-frequency characteristics. Its mechanism of action is to use high-frequency micro-vibration to disrupt the static friction state between the gravel or rock strata and the blade, reducing the sliding friction coefficient. At the same time, the high-frequency vibration can also generate micro-cracks on the rock surface, reducing the energy required for crushing.

[0069] For mixed soils, the controller performs layered adaptive control. When the cutting depth of the shovel is within the surface soil range, the current stiffness coefficient and damping coefficient corresponding to the soil type of the first soil layer are used. The controller monitors the changing trend of the equivalent cutting force in real time. When a step increase in the equivalent cutting force is detected and the increase exceeds a second threshold, the controller determines that the shovel has cut from the (x-1)th soil layer into the xth soil layer. A step increase means that within a short time (e.g., 0.2 seconds), the average value of the equivalent cutting force jumps from a relatively stable state to a significantly higher level, and this jump is not a spike caused by an instantaneous impact, but a continuous plateau-like increase. At this time, the controller transitions the current stiffness coefficient and damping coefficient to the stiffness coefficient and damping coefficient corresponding to the soil type of the xth soil layer within a preset time (x is a positive integer greater than 1 and less than or equal to N, where N is the total number of soil layers contained in the mixed soil). This smooth transition ensures that the opening change of the hydraulic valve is continuous, avoiding damage to the cylinder seals and pipelines caused by pressure spikes. Furthermore, during the transition of the shovel blade cutting from the (x-1)th soil layer into the xth soil layer, the controller superimposes an auxiliary second chatter signal onto the shovel blade. This second chatter signal is used to reduce the interlayer impact between different soil layers. The interface between different soil layers is usually the weakest structural surface but also the area with the most severe stress concentration. When the shovel blade cuts from a soft soil layer into a hard soil layer, the edge of the hard soil layer will generate a huge linear impact load on the shovel blade edge, which can easily cause the blade edge to crack or deform locally. The function of the second chatter signal is to pre-treat the interface with micro-vibration before the shovel blade touches the edge of the hard soil layer, so as to cause slight breakage or loosening at the edge of the hard soil layer, thereby reducing the instantaneous impact peak when the shovel blade crosses the interface.

[0070] In another embodiment of the invention, the method further includes the step of setting a virtual elastic layer above the target design elevation.

[0071] When a bulldozer performs a leveling operation, the ultimate goal is to level the ground to a fixed design elevation. In traditional control methods, the controller calculates the control input to the lifting cylinder using a PID algorithm based on the deviation between the current and target elevation of the blade. The problem with this method is that when the blade approaches the target elevation, the deviation is small, and the control output is also small. However, the blade still has some downward inertia and can easily exceed the target elevation, causing overcutting. The operator usually needs to manually intervene when approaching the target, significantly reducing the lowering speed and sacrificing operational efficiency.

[0072] To address this issue, the controller incorporates a virtual elastic layer with adjustable thickness above the target design elevation. This virtual elastic layer doesn't involve adding any physical structure; rather, it's a special control zone defined within the controller's algorithm logic, encompassing a distance range (e.g., 0 to 5 cm) above the target design elevation. Within this zone, the lifting cylinder's output force is no longer determined by conventional PID deviation control but is dominated by a virtual force-displacement relationship. The fourth stiffness coefficient of the virtual elastic layer is determined based on the current soil type: a smaller stiffness coefficient is used for loose backfill soil, allowing the blade to gently approach the target surface; a larger stiffness coefficient is used for dense silty clay, as dense soil is more sensitive to overcutting and requires a stronger pushing force.

[0073] Once the blade enters the virtual elastic layer, in positive stiffness mode, the controller adjusts the output force of the lifting cylinder based on the distance between the blade and the target design elevation, as well as the fourth stiffness coefficient, to maintain the contact force between the blade and the soil within a preset range. In positive stiffness mode, the output force of the lifting cylinder is negatively correlated with the distance between the blade and the target design elevation—the greater the distance, the smaller the thrust; the smaller the distance, the greater the thrust. This relationship contradicts conventional control intuition, but its physical effect is ingenious: as the blade rapidly approaches the target surface, the virtual elastic layer generates a gradually increasing reverse thrust, acting like an invisible air cushion to support the blade, slowing it down and ultimately allowing it to touch the target surface at almost zero speed, fundamentally eliminating the possibility of overcutting.

[0074] In one feasible embodiment, to address the problem of under-excavation (blade height exceeding target surface) in loose backfill soil, this invention introduces a negative stiffness mode. Specifically, when the current working soil type is loose backfill soil, and the current elevation of the blade is higher than the target design elevation, but the difference is less than a third threshold (e.g., 1 cm), the controller switches the virtual elastic layer from positive stiffness mode to negative stiffness mode. In negative stiffness mode, the lifting cylinder outputs a downward pulling force, causing the blade to be pulled towards the target design elevation. This is equivalent to switching a repulsive force field to an attractive force field, allowing the blade to actively pull towards the target surface instead of being pushed away. To suppress possible reciprocating oscillations in negative stiffness mode, the controller also adjusts the blade's damping coefficient to an overdamped state while switching to negative stiffness mode. In negative stiffness mode, when the blade elevation is lower than the target design elevation, the controller switches the virtual elastic layer back to positive stiffness mode, forming a complete hysteresis control logic.

[0075] Specifically, when the current working soil type is mixed soil, the controller prohibits the virtual elastic layer from entering negative stiffness mode. This is because the underlying layer of mixed soil may be extremely hard gravel or rock. If the blade is actively pulled towards the target surface in negative stiffness mode, and the target surface happens to be below hard rock, the blade may impact the rock with uncontrollable force, causing serious damage. This mutual exclusion logic reflects the invention's thorough consideration of safety boundaries.

[0076] In yet another embodiment, the method further includes a passive leveling step based on the difference between the left and right cutting forces.

[0077] When operating on transverse slopes, the left and right sides of a bulldozer blade often experience different cutting resistances. If the ground on the left is higher or the soil is harder, the cutting force on the left is greater than on the right, causing the blade to naturally tilt to the right, resulting in excessive cutting depth on the right and creating transverse unevenness. The traditional solution is to detect the transverse tilt angle of the blade using a tilt sensor and then actively adjust the extension and retraction of the left and right tilt cylinders to forcibly push the blade back to a horizontal position. This method has two inherent drawbacks: active adjustment requires overcoming significant soil resistance, resulting in high energy consumption and slow response; and in situations of high-speed travel or uneven soil conditions, closed-loop control is prone to oscillations, which can actually exacerbate unevenness.

[0078] In this embodiment, the controller first compares the difference in equivalent cutting forces between the left and right ends of the blade. When the difference exceeds a fourth threshold, it indicates that there is a significant lateral force imbalance in the blade. At this time, the controller does not attempt to actively adjust the cylinder to overcome this imbalance, but instead operates in the opposite direction: it identifies the side with less force (i.e., the side with less cutting resistance and shallower cutting depth) and then applies a third chatter signal to the tilting cylinder on that side.

[0079] The third chatter signal has the following characteristics: a frequency range of 5 to 50 Hz (too low a frequency will cause the blade to fluctuate, too high a frequency will not effectively reduce soil resistance); and a chatter amplitude of 5% to 15% of the rated pressure (too low an amplitude will not produce any effect, too high an amplitude may damage the cylinder). The mechanism of the third chatter signal is: high-frequency, micro-amplitude pressure pulsations cause minute separation and impact at the contact interface between the blade and the soil on that side, dynamically disrupting the soil's cohesion and friction angle, causing the resistance required for the blade to cut into the soil to drop instantaneously. Since the side with less force already has lower resistance, after the chatter further reduces the resistance, the blade on that side will automatically sink under its own gravity until the forces on both sides are rebalanced. The entire leveling action does not use any active position closed loop, relying entirely on physical gravity. Energy consumption is almost negligible, the response speed is limited by gravitational acceleration rather than the hydraulic response bandwidth, and there will be no oscillation.

[0080] In yet another embodiment, active compliance control also includes an impact-type soil-penetration strategy for hard strata.

[0081] For hard gravelly soils, rock layers, or hard substrata within mixed soils, even with the aforementioned high stiffness and high damping parameters and flutter assist, the incremental cutting depth of the shovel may still be very limited. In such cases, repeatedly implementing conventional active compliance control will only result in repeated bouncing without achieving substantial progress.

[0082] The controller maintains a counter to record the number of consecutive executions of active compliance control. When the current working soil type is hard gravelly soil or rock, or the underlying soil type of the mixed soil is hard gravelly soil or rock, and the cut-in depth increment is less than the fifth threshold after M consecutive executions of active compliance control cycles (M is a positive integer not less than 2, such as 3 times), the controller determines that the current formation hardness exceeds the processing capacity of conventional active compliance control.

[0083] At this point, the controller automatically exits the normal mode and switches to an impact-type soil entry strategy: First, it rapidly raises the blade to a preset impact starting height (e.g., 10 to 15 cm), which is significantly higher than the normal lifting height. Then, the controller executes a combined accelerated descent action: simultaneously releasing the lowering valve of the lifting cylinder and applying a momentary positive hydraulic impact force to its large chamber. Under the dual acceleration of gravity and hydraulic thrust, the blade achieves a speed far exceeding the normal descent speed, impacting the hard soil layer with greater momentum. The physical effect of this impact is to utilize the momentum theorem to generate a huge instantaneous impact force in a very short time, breaking the surface of the hard soil layer or creating a crack network in the dense structure, thus creating soil entry conditions for subsequent conventional cutting. After one impact, the controller immediately returns to the active compliant control mode, monitors the vehicle's attitude, and reassesses the cutting efficiency. If the cutting efficiency recovers after the impact, the controller returns to the normal mode; if it is still ineffective, the controller can execute the impact again or issue an alarm to prompt the operator that other construction methods need to be used in this area. This selective triggering mechanism ensures that the impact action is invoked only when the conventional method is truly ineffective, thus maximizing the lifespan of the equipment.

[0084] In this embodiment of the invention, the controller acquires attitude sensor data to predict the trend of vehicle pitch angle changes. When a sudden change in vehicle pitch angle is predicted to exceed a first threshold, active compliance control is executed. This prevents the blade impact from adversely affecting the vehicle's attitude and leveling accuracy, achieving intelligent compliance control of the blade. By further introducing multi-level control strategies such as soil type recognition, virtual elastic layer, flutter-assisted leveling, and impact-type soil entry, this invention can maintain stable operating performance across the entire working condition range, from loose backfill soil to hard rock layers, significantly improving the bulldozer's operating efficiency and accuracy.

[0085] In some embodiments, to improve stability in the negative stiffness mode, this disclosure provides a clear damping coefficient calibration range and a stability guarantee mechanism. The damping ratio ζ of a second-order system determines the system's response characteristics: when ζ < 1, it is an underdamped state, and the system response exhibits overshoot and oscillation; when ζ = 1, it is a critically damped state, and the system response has no overshoot but is critically stable; when ζ > 1, it is an overdamped state, and the system response has no overshoot and no oscillation, but the response speed is slow.

[0086] In negative stiffness mode, the system's equivalent stiffness is negative. Without proper damping compensation, the system will inevitably diverge. This embodiment compensates for the instability caused by negative stiffness by adjusting the damping coefficient to an overdamped state. Specifically, the controller sets the damping ratio ζ of the blade lifting cylinder control loop to a range greater than 1.2 and less than 2.0. If ζ is too low (e.g., 1.0 to 1.2), although the system is nominally overdamped, negative stiffness disturbances may cause it to enter the underdamped region, still posing a slight risk of oscillation. If ζ is too high (e.g., greater than 2.0), the system's response speed is too slow, and the time it takes for the blade to be pulled from a position above the target elevation to the target surface is too long, losing the rapid correction advantage of the negative stiffness mode.

[0087] For example, when the damping ratio ζ = F1, the negative stiffness mode can pull the blade from a position 1 cm above the target elevation to the target surface within 50 milliseconds without any overshoot or oscillation. When ζ = F3, the response time increases to 80 milliseconds but the stability is further improved; when ζ = F2, the response time is 40 milliseconds but there is a slight oscillation of about 0.3 mm (within the allowable error range of fine leveling operations). Therefore, the preferred damping ratio range represented by F2d is 1.3 to 1.7. F1 is less than F3 and greater than F2. Furthermore, the negative stiffness mode should only be used when the soil type is loose backfill. This is because the cutting resistance of loose backfill increases slowly with the cutting depth, and the downward pull generated by the negative stiffness mode can be effectively balanced by the soil resistance, forming a stable working point. In dense silty clay or hard gravelly soil, the soil resistance increases abruptly with the cutting depth. In the negative stiffness mode, once the blade is pulled into the soil layer, it may rebound due to the sudden increase in resistance, making it impossible to form a stable working point. Therefore, the negative stiffness mode should not be used in these soil types.

[0088] In some embodiments, this disclosure proposes a dual-mode switching control method, in which the controller runs two sets of control algorithms simultaneously: one is an impedance control mode, which treats the interaction between the shovel and the soil as a second-order system of mass, spring, and damping, and the output target of the controller is to make the shovel exhibit the desired mechanical impedance characteristics; the other is an admittance control mode, which takes the position change of the shovel as input and outputs the desired force, so that the shovel exhibits the characteristic of complying with the operating force.

[0089] The switching between the two control modes is managed by a mode scheduler. The mode scheduler evaluates the current operating stage and working condition characteristics in real time. Specifically, when the blade is in the free movement stage (i.e., not in contact with the soil or the cutting force is close to zero), the admittance control mode is used, enabling the blade to respond quickly to the operator's commands and reduce operating delays; when the blade begins to cut into the soil layer and the cutting force enters the stable range, it switches to the impedance control mode to maintain a stable cutting depth; when a rapid accumulation of elastic potential energy is detected and an impact is imminent, it switches back to the admittance control mode and reduces the system's expected impedance, allowing the blade to obediently avoid the impact direction and transform the destructive impact into a controllable yielding motion.

[0090] In one feasible embodiment, the mode scheduler employs a hysteresis switching strategy, where the cutting force threshold required to switch from admittance mode to impedance mode is higher than the cutting force threshold required to switch back from impedance mode to admittance mode. This hysteresis characteristic prevents frequent mode oscillations near the critical cutting force value, ensuring system stability.

[0091] In the process of proactive adaptive control execution, the mode scheduler plays a key role in prediction, switching, and recovery.

[0092] Specifically, when the elastic potential energy observer predicts that the elastic potential energy will exceed the first energy threshold within a certain period of time (e.g., 0.1 seconds), the mode scheduler does not simply trigger active compliant control, but first performs a mode switch: switching the current control mode from impedance mode (or normal position mode) to admittance control mode, while reducing the system's expected impedance value to 20% to 30% of the normal value. The physical effect of this switch is that the blade control system transforms from a rigid position holder into a compliant mechanical spring. When an impact occurs, the blade does not resist it directly but actively yields, converting the destructive impact energy into a controllable yield displacement.

[0093] After the impact energy is released in admittance mode, the mode scheduler monitors the recovery of the vehicle's pitch angle. When it detects that the pitch angle has recovered to more than 90% of its pre-abrupt state, the mode scheduler does not immediately switch back to impedance mode, but instead enters a transition state. In the transition state, the controller uses hybrid control parameters between impedance and admittance modes, allowing the blade to re-contact the soil with controlled compliance, avoiding secondary impacts caused by rigid reset. Only after the blade re-establishes a stable cutting state (indicated by the fluctuation amplitude of the equivalent cutting force being less than a preset threshold) does the mode scheduler switch the system back to impedance mode.

[0094] In one feasible embodiment, the mode scheduler also dynamically adjusts the desired impedance value in the admittance mode based on the output of the aforementioned soil type classifier. For loose backfill soil, a lower desired impedance value is set to make the blade exhibit highly compliant characteristics, avoiding triggering frequent active compliance due to slight disturbances; for hard gravelly soil or rock strata, a higher desired impedance value is set to enable the blade to quickly recover cutting force after active compliance, avoiding excessive blade retraction that leads to decreased work efficiency. This soil type adaptive desired impedance adjustment is a deep integration of dual-mode switching control and the aforementioned soil type recognition technology.

[0095] In one feasible embodiment, when the blade enters the virtual elastic layer (i.e., the difference between the blade elevation and the target design elevation is less than the thickness of the virtual elastic layer), the mode scheduler switches the control system from impedance mode to admittance mode. The logic behind this switch is that in impedance mode, the controller attempts to maintain the blade's position, which contradicts the goal of the virtual elastic layer to allow the blade to gently approach the target; while in admittance mode, the controller yields to external forces (including virtual forces generated by the virtual elastic layer), allowing the blade to smoothly follow the force-displacement relationship of the virtual elastic layer.

[0096] In admittance mode, the controller takes the virtual force generated by the virtual elastic layer as input and outputs the desired blade movement speed. Specifically, the force-displacement relationship of the virtual elastic layer is determined by the aforementioned fourth stiffness coefficient. When the virtual elastic layer is in positive stiffness mode, the virtual force exhibits a repulsive force, which the admittance controller converts into an upward movement speed, causing the blade to decelerate and approach the target. When the virtual elastic layer switches to negative stiffness mode, the virtual force exhibits an attractive force, which the admittance controller converts into a downward movement speed, causing the blade to be actively pulled towards the target.

[0097] There is also a priority coordination mechanism between the mode scheduler and the virtual elastic layer mode switching logic. As mentioned earlier, when the soil type is mixed soil, the virtual elastic layer is prohibited from entering the negative stiffness mode. At the same time, the mode scheduler also adjusts accordingly: after entering the virtual elastic layer in mixed soil, the mode scheduler prohibits switching from the impedance mode to the admittance mode, and instead adopts a force-position hybrid control in the impedance mode—that is, while maintaining the position closed loop, the virtual force of the virtual elastic layer is superimposed on the control output as a feedforward. This coordination mechanism ensures that the blade always maintains sufficient position stiffness in uncertain mixed soil, avoiding uncontrolled downward thrust due to excessive compliance.

[0098] In one feasible embodiment, in a lateral leveling scenario, the aforementioned gravity-assisted passive leveling technology applies a third chatter signal to the tilting cylinder on the side with less force, causing the blade on that side to naturally sink under gravity. Dual-mode switching control can further enhance the effectiveness of this passive leveling technology.

[0099] Specifically, when the controller detects that the difference in equivalent cutting forces between the left and right sides of the blade exceeds the fourth threshold, the mode scheduler does not immediately trigger gravity-assisted leveling. Instead, it first switches the blade's main control mode (i.e., the lifting cylinder control circuit that controls the blade's elevation) from impedance mode to admittance mode. The purpose of this switch is that during gravity-assisted leveling, the overall elevation of the blade may change slightly due to one side sinking. If elevation control is performed in impedance mode, the controller will attempt to maintain the blade's elevation unchanged, thereby counteracting the elevation change caused by tilting adjustments, which would hinder the leveling process.

[0100] After switching to admittance mode, the blade's elevation control loop becomes smoother, allowing for natural, minor elevation changes during tilt adjustments. At this point, the controller applies a third chatter signal to the tilt cylinder on the side with less force, causing that side to sink. Once the forces on both sides are balanced, the mode scheduler then determines, based on the current elevation deviation, whether to maintain admittance mode (if the elevation deviation is within acceptable limits) or switch back to impedance mode (if precise elevation recovery is required).

[0101] In one feasible embodiment, the mode scheduler also determines the compliance of the admittance mode during gravity-assisted leveling based on the output of the aforementioned soil type classifier. For loose backfill soil, the desired impedance value of the admittance mode is set lower, allowing the blade elevation to freely change with tilt adjustments; for dense silty clay, the desired impedance value is set to a medium level, allowing the blade elevation to maintain basic stability while allowing some variation; for hard gravelly soil or rock strata, switching to admittance mode is prohibited during gravity-assisted leveling, and instead, the traditional active leveling method is used, because small changes in blade elevation in hard strata may lead to unintended contact between the cutting edge and hard points.

[0102] In one feasible embodiment, dual-mode switching control also plays a crucial role in the execution of the impact-type soil entry strategy. As mentioned earlier, the impact-type soil entry strategy involves raising the blade to a preset height after multiple attempts at active compliance control fail, and then using a combination of gravity and hydraulic thrust to accelerate its descent, impacting the hard soil layer with significant momentum. The core challenge of this strategy is that the impact generates a huge reverse impact force, which, if left uncontrolled, could damage the vehicle body and hydraulic system.

[0103] This invention combines dual-mode switching control with an impact-based soil-penetrating strategy, switching modes at the instant of impact to achieve impact buffering. Specifically, the controller accurately predicts the timing of the blade's contact with the soil (based on the descent speed and current height). One instant before contact (e.g., 20 to 50 milliseconds prior), the mode scheduler switches the control system from the current impedance mode (or position mode) to admittance mode, setting the desired impedance value to an extremely low value. This switch allows the blade to exhibit high compliance upon impact with hard soil layers, converting impact energy into blade yield displacement rather than transferring it to the vehicle body.

[0104] After impact, the mode scheduler monitors the blade's rebound. If the blade rebounds significantly after impact (detected by the displacement sensor), the mode scheduler maintains the admittance mode for a period, allowing the blade to fall back naturally under gravity. During the fall, the desired impedance value is gradually increased to ensure the blade smoothly re-adheres to and cuts into the soil. If the blade does not rebound significantly after impact (indicating successful cutting), the mode scheduler immediately switches back to impedance mode to maintain the achieved cutting depth.

[0105] In one feasible embodiment, the mode scheduler also records the optimal retreat distance during each impact—that is, the maximum displacement of the blade in admittance mode at the moment of impact. This retreat distance is negatively correlated with the stiffness of the hard soil layer: the harder the soil layer, the smaller the retreat distance. The controller uses this relationship to estimate the equivalent stiffness of the hard soil layer online and uses it to optimize the target momentum value for the next impact: if the retreat distance is too small (indicating that the soil layer is extremely hard), the target momentum is increased; if the retreat distance is too large (indicating that the soil layer is not as hard as expected), the target momentum is decreased to avoid excessive impact.

[0106] In one feasible embodiment, the mode scheduler employs a hysteresis switching strategy to prevent mode oscillations. The specific implementation of hysteresis switching is as follows: The cutting force threshold required to switch from admittance mode to impedance mode is set as a second threshold F_th2, while the cutting force threshold required to switch back from impedance mode to admittance mode is set as a third threshold F_th3, where F_th2 is greater than F_th3. The difference between these two values ​​constitutes the hysteresis interval.

[0107] For example, in a scenario where a shovel cuts into the soil: when the shovel is in free fall (cutting force is close to zero), the system is in admittance mode. As the shovel contacts the soil, the cutting force gradually increases. When the cutting force exceeds F_th2 (e.g., 5000 Newtons), the mode scheduler switches the system to impedance mode, allowing the shovel to maintain a stable cutting depth. If the soil softens locally and the cutting force decreases, the system remains in impedance mode as long as the cutting force is still above F_th3 (e.g., 3000 Newtons) and does not immediately switch back to admittance mode. Only when the cutting force truly drops below F_th3 (e.g., when the shovel is lifted off the soil) does the system switch back to admittance mode.

[0108] When the cutting force fluctuates between F_th3 and F_th2 (e.g., around 4000 Newtons), the system avoids frequent mode switching, thus preventing mode oscillations from impacting the hydraulic valves and control algorithm. The width of the hysteresis interval (F_th2 - F_th3) can be dynamically adjusted according to soil type: for loose backfill soil, due to larger cutting force fluctuations, a wider hysteresis interval is set (e.g., 2500 Newtons); for dense silty clay, due to relatively stable cutting force, a narrower hysteresis interval is set (e.g., 1000 Newtons). This adaptive hysteresis width represents another deep integration of the mode scheduler with the aforementioned soil type recognition technology.

[0109] In a more preferred embodiment, the mode scheduler also uses the output of the aforementioned elastic potential energy observer as an auxiliary basis for mode switching.

[0110] Specifically, even if the current cutting force has not yet reached F_th2, if the elastic potential energy E_elastic estimated by the elastic potential energy observer exceeds the preset potential energy threshold E_th, the mode scheduler will switch the system from impedance mode to admittance mode in advance. Thus, under certain operating conditions (such as when there is a buried rock edge in front of the blade), the increase in cutting force can be very rapid, with only a few tens of milliseconds between falling below F_th2 and the impact occurring. If the switch is delayed until the cutting force exceeds F_th2, the impact may occur before the switch is completed due to the response delay of the hydraulic system and control algorithm.

[0111] By introducing the predictive signal from the elastic potential energy observer, the mode scheduler can complete the mode switching earlier before the impact occurs, so that the blade is already in the compliant state of the admittance mode when the impact arrives.

[0112] In some embodiments, such as Figure 4 As shown, this embodiment of the disclosure provides a bulldozer blade control device, including: Prediction module 4110 is used to predict the trend of the bulldozer's body pitch angle based on attitude sensing data; The execution module 4120 is used to perform active compliance control when the positive abrupt change in the vehicle body pitch angle is predicted to exceed a first threshold based on the trend of the vehicle body pitch angle change; wherein, the positive abrupt change in the vehicle body pitch angle is the change in the sinking of the bulldozer's front end relative to the bulldozer's body within a preset time slice; the active compliance control includes: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, so as to release the elastic potential energy accumulated on the blade's cutting edge; during the process of the vehicle body pitch angle falling back under the action of gravity, using the falling inertia and the blade falling command to control the blade so that the blade re-contacts the soil layer with a preset cutting force.

[0113] Figure 3B The diagram illustrates a scenario of multiple bulldozers working collaboratively. When multiple bulldozers operate side-by-side (e.g., for large-scale site leveling) or in relay (e.g., for slope bulldozing), the impact of a single bulldozer's blade not only affects its own posture but also transmits through the soil to adjacent or subsequent bulldozers, causing a chain reaction. In existing technologies, each bulldozer independently performs active compliant control, lacking coordination.

[0114] Establish a shared elastic potential energy field model among multiple bulldozers. Each bulldozer broadcasts its perceived equivalent cutting force of the blade, the trend of the vehicle's pitch angle, and the predicted direction and intensity of elastic potential energy release to other bulldozers in the work area in real time via vehicle-to-everything (V2X) communication.

[0115] The controller receives elastic potential energy prediction data from neighboring bulldozers and calculates a spatiotemporal attenuation model of this elastic potential energy propagating through the soil layer to the controller. The propagation delay depends on the soil type (obtained through the aforementioned soil type identification) and the propagation distance; the attenuation magnitude depends on the soil damping characteristics.

[0116] When it is predicted that a neighboring bulldozer will release elastic potential energy (i.e., cause an impact) at a certain moment, and that the impact will cause an observable disturbance to the attitude of the machine after propagating through the soil, the machine will perform pre-synchronization compliant control: within a preset time window before the arrival of the shock wave, the stiffness and damping of the machine's blade will be actively adjusted to make it compliant, so as to absorb the propagated disturbance energy rather than resist it rigidly.

[0117] Furthermore, multiple bulldozers can form a virtual formation, with one acting as the navigator. When the navigator detects a hard obstacle ahead, it not only performs its own active compliant control but also sends avoidance timing instructions to other bulldozers in the formation. This ensures that the lifting and lowering actions of each bulldozer are staggered according to time differences, avoiding a sharp drop in work efficiency caused by multiple bulldozers lifting their blades simultaneously, and also avoiding soil resonance caused by multiple bulldozers lowering their blades simultaneously.

[0118] In this way, the active compliance control of a single machine is extended to the disturbance isolation and energy absorption of multiple machines in a coordinated manner, which prevents disturbances from propagating and amplifying in the fleet and reduces the overall attitude overshoot of the fleet. In some situations, when multiple bulldozers work collaboratively on the same site, the ground needs to be leveled to the same target design elevation. Due to differences in blade wear, hydraulic system characteristics, and uneven soil layers encountered by each bulldozer, even when executing the same control commands, the actual working results of each machine will vary, resulting in seams or steps on the leveled ground. In this embodiment, a distributed consistency control framework for blade trajectory is established. Each bulldozer records its own actual working elevation trajectory in real time and shares it with neighboring bulldozers through a communication network. The controller no longer relies solely on the absolute target design elevation but also references the actual working elevation of neighboring bulldozers, executing a consistency error convergence algorithm.

[0119] Specifically, the expected blade elevation of each bulldozer is composed of three weighted components: the target design elevation (global reference), the average actual elevation of neighboring bulldozers (local reference), and the bulldozer's historical operating elevation (time reference). The weighting coefficients are dynamically adjusted based on the spatial relationship between bulldozers: adjacent bulldozers with high operational overlap have higher weights; bulldozers that are far apart or have no overlap have lower weights.

[0120] When the deviation between the actual elevation of the machine and the average actual elevation of its neighbors exceeds a threshold, the controller adjusts the cutting force of the blade and the stiffness coefficient of the virtual elastic layer to gradually converge the elevation of the machine towards that of its neighbors. The convergence process uses a gradual adjustment to avoid creating new unevenness due to rapid correction.

[0121] Furthermore, a blade wear compensation model is introduced. Each bulldozer periodically (or after completing a work segment) performs calibration on a standard test surface to measure the actual wear of the blade edge, and shares the wear data with the entire fleet. Bulldozers with more severe wear receive higher weight in consistency control, causing other bulldozers to gravitate towards them rather than the other way around, thus preventing severely worn blades from becoming isolated information silos.

[0122] In this way, the consistency of ground elevation is significantly improved after multi-machine operation, and eliminating differences in splicing seams, steps, or blade wear will not lead to a decline in work quality.

[0123] In some embodiments, during large-scale site leveling operations, the flow of earthwork within the site is complex when multiple bulldozers operate simultaneously. If each bulldozer operates independently, some areas may experience repeated bulldozing while others suffer from insufficient earthwork, leading to overall inefficiency. Therefore, a multi-machine collaborative framework for balancing earthwork flow is established. A digital elevation model (DEM) of the work area is acquired in real time using a drone or fixed laser scanner. Combined with the position, speed, and blade posture of each bulldozer, the actual earthmoving volume (the volume of earth moved per unit time) of each bulldozer is calculated.

[0124] The controller divides the entire work area into several virtual grids, each with the goal of achieving the designed elevation after the work is completed. Each bulldozer's task is not to cover a fixed area, but rather to adjust dynamically: when the current elevation of a grid is significantly higher than the design value, that grid is marked as a soil supply area; when the elevation of a grid is significantly lower than the design value, that grid is marked as a soil demand area.

[0125] The bulldozers obtain a map of the supply and demand areas via V2X communication. The controller plans the earthmoving route for each bulldozer: taking soil from the nearest supply area and pushing it to the nearest demand area for unloading. The route planning also considers the bulldozer's turning energy consumption, pushing distance, and path conflicts with other bulldozers.

[0126] When multiple bulldozers simultaneously transport soil to the same area requiring soil, the controller executes a sequential unloading strategy: based on the estimated arrival time of each bulldozer, it automatically arranges the unloading order and instructs waiting bulldozers to temporarily store in the buffer zone or slow down, avoiding congestion caused by multiple bulldozers arriving at the same time.

[0127] When the blade impact-type soil-penetrating strategy is triggered (e.g., encountering hard rock), the bulldozer reports the impact event and location to the controller. The controller temporarily marks the area as a "pending zone" and dispatches other bulldozers to bypass it, avoiding repeated impacts in the same hard area by multiple bulldozers, thus preventing wasted efficiency. After a successful impact, the area returns to normal operating conditions.

[0128] This achieves dynamic balance of earthwork within the site, reduces empty runs and ineffective transport, and makes the treatment of hard areas more focused and efficient. In some embodiments, the aforementioned single-machine scheme employs flutter signals (first, second, and third flutter signals) to reduce soil resistance. When multiple bulldozers operate simultaneously in the same site and are close together, the flutter signals generated by each machine may propagate through the soil and interfere with each other. If the flutter frequencies of two bulldozers are the same or close, resonance amplification may occur, causing unexpected impacts to the vehicle body; if the frequencies differ slightly, beat frequency oscillation may occur, making the blade control unstable. In this embodiment, a multi-machine collaborative mechanism with orthogonalized flutter frequencies is established. Before activating the flutter function, each bulldozer obtains the flutter frequencies currently used by other bulldozers in the work area via V2X communication. The controller uses a frequency orthogonal allocation algorithm to assign a frequency value to its own flutter signal that is not an integer multiple of the frequencies of all its neighbors.

[0129] Specifically, the available frequency range (e.g., 5Hz to 50Hz) is divided into several orthogonal channels, with the center frequencies of the channels satisfying a non-harmonic relationship (i.e., the ratio of any two frequencies is not a rational number, or at least not a small integer ratio, such as 1:2, 2:3, 3:4, etc.). The controller acts as a virtual frequency scheduler, dynamically allocating orthogonal channels to each bulldozer. When a bulldozer leaves the work area or stops fluttering, the frequency it occupies is released and made available for use by other bulldozers.

[0130] In situations where frequency conflicts cannot be completely avoided (e.g., the number of bulldozers exceeds the number of orthogonal channels), the controller employs a time-division multiplexing strategy: time is divided into multiple time slots, and each bulldozer applies a flutter signal within its own time slot, while other bulldozers pause flutter or switch to a low-amplitude mode within that time slot. The time slot allocation can be dynamically adjusted based on the operational priority of each bulldozer or the current soil hardness encountered.

[0131] Furthermore, for bulldozers operating in hard gravel soil or rock formations that require continuous application of the first flutter signal, the controller assigns a dedicated frequency to it and notifies other bulldozers to avoid using that frequency or its second or third harmonics within a certain distance (e.g., 20 meters) from the bulldozer.

[0132] This eliminates the risk of frequency interference and resonance from multi-machine flutter, ensures that the flutter and drag reduction effect of each bulldozer is not affected by neighboring machines, and improves the safety of close-range collaborative operation of multiple machines. In some cases, the virtual elastic layer in a single-machine scheme involves setting a virtual force field above the target design elevation to allow the blade to smoothly approach the target surface. In multi-machine collaborative leveling operations, as bulldozers successively pass through the same area, the actual elevation after the first bulldozer's operation may be slightly higher or lower than the design value. If the virtual elastic layer of the second bulldozer still targets the original design elevation, it may result in overcutting or undercutting. Therefore, a multi-machine collaborative mechanism with a dynamic virtual elastic layer is established. After each bulldozer completes its operation, it uploads its actual operating elevation (accurate to the centimeter level) to a cloud platform or broadcasts it to subsequent bulldozers. The virtual elastic layer of subsequent bulldozers no longer targets the fixed original design elevation but instead targets a dynamic target elevation. The dynamic target elevation comprehensively considers the original design elevation and the actual operating results of the preceding bulldozers, using a weighted average or Kalman filter for fusion estimation.

[0133] Specifically, for areas where no bulldozers have yet operated, the dynamic target elevation is equal to the original design elevation, and the thickness and stiffness of the virtual elastic layer are set normally. For areas where one or more bulldozers have already operated, the dynamic target elevation is adjusted based on the results of previous operations: if the results of previous operations are generally higher than the design value, the dynamic target elevation is appropriately moved upward to avoid excessive downcutting by subsequent bulldozers; if the results of previous operations are generally lower than the design value, the dynamic target elevation is appropriately moved downward to avoid missed excavations and the formation of raised areas by subsequent bulldozers.

[0134] The stiffness coefficient of the virtual elastic layer is also adjusted according to the variance of the results of the preceding operations: if the consistency of the results of the preceding operations is high (small variance), the stiffness coefficient is appropriately increased to make the blade follow the dynamic target elevation more accurately; if the consistency of the results of the preceding operations is low (large variance), the stiffness coefficient is appropriately decreased to make the blade adapt more flexibly to the local terrain.

[0135] Furthermore, a learning mechanism for a virtual elastic layer is introduced. The controller records the deviation between the actual elevation and the dynamic target elevation after each operation, and updates the fusion weights of the dynamic target elevation through an online learning algorithm (such as recursive least squares), so that the deviation in subsequent operations gradually converges. After several operations, the entire fleet can "remember" the terrain features of the site and realize an adaptive fine-leveling strategy.

[0136] In this way, the flatness of the ground is significantly improved after multiple machines work in sequence, reducing the need for repeated corrections due to errors in previous operations, and enabling the machine group to collectively learn and memorize the site.

[0137] The bulldozer blade control device provided in this embodiment is used to execute the bulldozer blade control method of any of the foregoing embodiments.

[0138] like Figure 4 As shown, a bulldozer blade control device includes: Prediction module 4110 is used to predict the trend of the bulldozer's body pitch angle based on attitude sensing data; The execution module 4120 is used to perform active compliance control when the positive abrupt change in the vehicle body pitch angle is predicted to exceed a first threshold based on the trend of the vehicle body pitch angle change; wherein, the positive abrupt change in the vehicle body pitch angle is the change in the sinking of the bulldozer's front end relative to the bulldozer's body within a preset time slice; the active compliance control includes: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, so as to release the elastic potential energy accumulated on the blade's cutting edge; during the process of the vehicle body pitch angle falling back under the action of gravity, using the falling inertia and the blade falling command to control the blade so that the blade re-contacts the soil layer with a preset cutting force.

[0139] Combination Figure 5 As shown, this application embodiment provides an electronic device, which may include a processor 10 and a memory 11. Optionally, the electronic device may further include a communication interface 12 and a bus 9. The processor 10, communication interface 12, and memory 11 can communicate with each other via the bus 9. The communication interface 12 can be used for information transmission. The processor 10 can call logical instructions in the memory 11 to execute the bulldozer blade control method of the above embodiment.

[0140] Furthermore, the logical instructions in the aforementioned memory 11 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0141] The memory 11, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this application. The processor 10 executes functional applications and data processing by running the program instructions / modules stored in the memory 11, thereby implementing the bulldozer blade control method in the above embodiments.

[0142] The memory 11 may include a program storage area and a data storage area. The program storage area may store the operating system and application programs required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 11 may include high-speed random access memory and may also include non-volatile memory.

[0143] The embodiments or examples disclosed in this application are not exhaustive, but merely illustrative of some embodiments or examples, and are not intended to limit the scope of protection of this disclosure. Unless contradictory, each step in a particular embodiment or example can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment or example can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment or example can be arbitrarily interchanged. Furthermore, optional methods or examples in a particular embodiment or example can be arbitrarily combined; moreover, embodiments or examples can be arbitrarily combined. For example, some or all steps of different embodiments or examples can be arbitrarily combined, and a particular embodiment or example can be arbitrarily combined with optional methods or examples of other embodiments or examples.

[0144] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0145] 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 units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0146] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0147] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or 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 of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0148] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for controlling the blade of a bulldozer, characterized in that, include: The trend of the bulldozer's body pitch angle change is predicted based on attitude sensing data; When the positive abrupt change in the vehicle body pitch angle is predicted to exceed a first threshold based on the trend of the vehicle body pitch angle change, active compliance control is executed; wherein, the positive abrupt change in the vehicle body pitch angle is the change in the sinking of the bulldozer's front end relative to the bulldozer's body within a preset time slice; the active compliance control includes: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, so as to release the elastic potential energy accumulated on the blade's cutting edge; during the process of the vehicle body pitch angle falling back under the action of gravity, the blade is controlled by the falling inertia and the blade falling command to make the blade re-contact the soil layer with a preset cutting force.

2. The method according to claim 1, characterized in that, Before the active compliance control is executed, the method further includes: Collect pressure data from the hydraulic system of the shovel. Calculate the equivalent cutting force on the shovel based on the pressure data; Feature vectors are extracted based on the time-series data of the equivalent cutting force, wherein the feature vectors include at least one of the mean, variance, peak frequency, and impact energy of the equivalent cutting force. The feature vector is input into a classifier to identify the current soil type, wherein the current soil type is loose backfill soil, dense silty clay, hard gravelly soil or rock layer, or mixed soil; the mixed soil includes different types of soil layers; Adjust and control the stiffness coefficient and damping coefficient of the shovel blade according to the current soil type.

3. The method according to claim 2, characterized in that, The step of adjusting and controlling the stiffness coefficient and damping coefficient of the shovel blade according to the current soil type includes: The blade is controlled using a first stiffness coefficient and a first damping parameter for the loose backfill soil. The blade is controlled using a second stiffness coefficient and a second damping parameter for the dense silty clay, wherein the second stiffness coefficient is higher than the first stiffness coefficient; and the second damping parameter is higher than the first damping parameter. For the hard gravelly soil or rock strata, a third stiffness coefficient and a third damping parameter are used, and a first flutter signal is superimposed to control the blade; the third stiffness coefficient is higher than the second stiffness coefficient; the third damping parameter is higher than the second damping parameter; the frequency of the first flutter signal is higher than a preset frequency; For the mixed soil, stratified adaptive control is performed, including: When the cutting depth of the shovel is within the surface soil range, the current stiffness coefficient and the current damping coefficient corresponding to the first soil type of the mixed soil are used. When a step increase in the equivalent cutting force is detected and the increase exceeds the second threshold, it is determined that the shovel cuts from the (x-1)th soil layer into the xth soil layer. Within a preset time, the current stiffness coefficient and the current damping coefficient are transitioned to stiffness coefficients and damping coefficients corresponding to the soil type of the xth soil layer; x is a positive integer greater than 1 and less than or equal to N; N is the total number of soil layers contained in the mixed soil. During the transition from the (x-1)th soil layer to the xth soil layer, a second flutter signal is superimposed on the shovel; the second flutter signal is used to reduce the interlayer impact between different soil layers.

4. The method according to claim 2 or 3, characterized in that, The method further includes: An adjustable-thickness virtual elastic layer is set above the target design elevation; the fourth stiffness coefficient of the virtual elastic layer is determined according to the current soil type. When the shovel blade enters the virtual elastic layer, and the virtual elastic layer is in positive stiffness mode, the output force of the lifting cylinder of the shovel blade is controlled according to the distance between the shovel blade and the target design elevation and the fourth stiffness coefficient, so that the contact force between the shovel blade and the soil is maintained within a preset range, wherein the output force of the lifting cylinder is negatively correlated with the distance.

5. The method according to claim 4, characterized in that, The method further includes: When the current working soil type is loose backfill soil and the current elevation of the shovel is higher than the target design elevation and the difference is less than the third threshold, the virtual elastic layer switches from the positive stiffness mode to the negative stiffness mode. When the current working soil type is loose backfill soil and the blade elevation is lower than the target design elevation, the virtual elastic layer switches back to the positive stiffness mode; When the current working soil type is mixed soil, the virtual elastic layer is prohibited from entering the negative stiffness mode; In the positive stiffness mode, the lifting cylinder outputs an upward thrust to prevent the blade from approaching the target design elevation; and / or, In the negative stiffness mode, the lifting cylinder outputs a downward pulling force, causing the blade to be pulled towards the target design elevation; and / or, In the negative stiffness mode, the damping coefficient of the blade is adjusted to an overdamped state.

6. The method according to claim 2 or 3, characterized in that, The method further includes: Compare the difference in equivalent cutting forces at the left and right ends of the shovel; When the difference exceeds the fourth threshold, a third chatter signal is applied to the tilting cylinder on the side with less force to reduce the soil cutting resistance, so that the shovel automatically sinks under the action of gravity to achieve passive leveling.

7. The method according to claim 6, characterized in that, The third flutter signal has at least one of the following characteristics: The frequency range is 5 to 50 Hz; The flutter amplitude is 5% to 15% of the rated pressure.

8. The method according to claim 2 or 3, characterized in that, The active compliance control also includes: When the current working soil type is hard gravelly soil or rock layer, or the lower soil type of the mixed soil is hard gravelly soil or rock layer, and the cutting depth increment is less than the fifth threshold after M consecutive active compliance control cycles, the blade is raised to a preset height, and then the blade is lowered by a combination of gravity and hydraulic thrust; M is a positive integer not less than 2.

9. A bulldozer blade control device, characterized in that, include: The prediction module is used to predict the trend of the bulldozer's body pitch angle based on attitude sensing data. An execution module is used to perform active compliance control when a positive abrupt change in the vehicle body pitch angle is predicted to exceed a first threshold based on the trend of the vehicle body pitch angle change; wherein, the positive abrupt change in the vehicle body pitch angle is the change in the sinking of the bulldozer's front end relative to the bulldozer's body within a preset time slice; the active compliance control includes: sending a lifting command to the lifting cylinder of the blade to raise the blade upward, so as to release the elastic potential energy accumulated on the blade's cutting edge; during the process of the vehicle body pitch angle falling back under the action of gravity, using the falling inertia and the blade falling command to control the blade so that the blade re-contacts the soil layer with a preset cutting force.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions; when executed by a processor, the computer-executable instructions are able to implement the method described in any one of claims 1 to 8.

Images