An adaptive control method and system for grinding depth of an automatic floor grinder

Abstract

This invention discloses an adaptive control method and system for the grinding depth of an automatic floor grinder, belonging to the technical field of automatic / intelligent construction and flooring construction equipment. The method includes the following steps: S1, data acquisition and preprocessing; S2, real-time grinding depth calculation; S3, dynamic adjustment of travel speed: obtaining the system-preset target grinding depth d_target, calculating the depth deviation value e=d_target-d_current, calculating the grinding machine travel speed adjustment amount ΔV through a PID control algorithm; calculating the target travel speed V_target, issuing speed commands to adjust the grinding machine's travel speed, and dynamically adapting the grinding time according to the depth deviation to make the actual grinding depth approach the target grinding depth; S4, repeated iteration: cyclically executing steps S1-S3 to form a closed-loop adaptive control. This invention achieves adaptive closed-loop control, improves construction quality and consistency, reduces hardware costs and system complexity, has a high degree of intelligence, and strong model scalability.

Description

Technical Field

[0001] This invention belongs to the technical field of automatic / intelligent construction and flooring equipment, and particularly relates to an adaptive control method and system for the grinding depth of an automatic floor grinding machine. Background Technology

[0002] Floor grinding machines are key equipment used for surface treatment of concrete floors, epoxy floors, etc. Their working principle is to use a grinding motor to drive the grinding disc to rotate, which causes the grinding pads attached to it to rub against the ground at high speed, thereby achieving processes such as leveling and polishing.

[0003] Existing remote-controlled or hand-push floor grinders rely primarily on operator experience to control grinding depth. At a fixed grinding stage (e.g., using a fixed grit size grinding disc and a fixed counterweight), the operator observes the floor's flatness and adjusts the grinder's speed in real-time. Slowing down the speed increases the grinding time in the same area, resulting in a greater grinding depth, and vice versa. While this method can meet basic flatness requirements, it demands extremely high levels of operator experience and physical strength, and results in inconsistent workmanship.

[0004] With the development of automatic navigation technology, automatic floor grinding machines capable of automatically scanning, mapping, and planning routes have emerged. However, when performing multi-stage grinding tasks, existing automatic grinding machines typically use a preset fixed speed parameter during a certain stage of grinding. Unlike humans, the machines cannot perceive minute changes in surface bumps in real time and adjust their speed accordingly. This leads to a problem: when encountering locally high bumps, the fixed-parameter machine cannot perform "focused grinding," leaving residual bumps and affecting the overall flatness of the final surface. Furthermore, relying solely on preset motor speed, counterweight, and grinding disc grit cannot form a closed-loop, adaptive depth control system. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides an adaptive control method and system for the grinding depth of an automatic floor grinder. It has the advantages of achieving adaptive closed-loop control, improving construction quality and consistency, reducing hardware costs and system complexity, high level of intelligence, and strong model scalability. It solves the problems of low automation, fixed travel speed that cannot adapt to uneven ground conditions, resulting in poor grinding depth control accuracy, poor construction quality consistency, and high hardware cost for high-precision modification in the prior art.

[0006] This invention is implemented as follows: an adaptive control method for the grinding depth of an automatic floor grinder includes the following steps:

[0007] Core mathematical model (soft measurement model of grinding depth)

[0008] The load torque (T) of the grinding motor during stable operation is related to the grinding power (P) and speed (N) by the following relationship: P = T * N. The grinding power is mainly consumed to overcome the friction between the grinding disc and the ground. This friction is related to the normal force (F_n) of the grinding disc on the ground, the coefficient of friction (μ), the contact area between the grinding disc and the ground, and the relative speed. In automatic grinding machines, the grinding depth (d) directly affects the normal force and the coefficient of friction.

[0009] Therefore, we propose the following functional relationship: T (or I) = f(N,G,C,d), where:

[0010] T: Real-time torque of the grinding motor (or I: real-time current, which is approximately proportional to the torque).

[0011] N: Real-time rotational speed of the grinding motor (rpm)

[0012] G: Current grinding disc mesh size (dimensionless, the larger the value, the finer the grinding disc).

[0013] C: Ground material coefficient (experimental calibration value, such as 1.0 for ordinary concrete and 1.3 for high-hardness emery flooring, etc.).

[0014] d: Real-time grinding depth (mm)

[0015] The formula for calculating the grinding depth (d) is as follows: Through extensive experiments on a specific type of grinding machine, the inverse function of the above function can be fitted, namely: d = F_inv(T, N, G, C)

[0016] A simplified engineering implementation model can be expressed as: d = K0*(T / N)*(1 / G)*C+d0

[0017] K0: System calibration coefficient (constant) related to the equipment and grinding discs.

[0018] d0: Initial offset (zero point when the grinding disc just contacts the ground but has not yet created depth)

[0019] More precisely, considering the nonlinear characteristics of the grinding process, a piecewise function or a machine learning model (such as a neural network) can be used: d=NN_model(T,N,G,C), where NN_model is a pre-trained shallow neural network (input layer: T,N,G,C; output layer: d), and the training data comes from calibration experiments under different working conditions;

[0020] Control method flow (based on the above mathematical model):

[0021] S1. Data Acquisition and Preprocessing

[0022] Complete the initialization of system parameters. The control system collects the real-time torque T or working current I and real-time rotational speed N of the grinding motor in real-time at a high frequency (such as 100 Hz), reads the number of mesh of the grinding disc G set by the current system, and the ground material coefficient C selected through the human-machine interface;

[0023] S2. Real-time grinding depth calculation

[0024] Substitute the collected I (or T), N, G, C into the above formula d = F_inv(T, N, G, C) or the trained neural network model NN_model to calculate the current grinding depth d_current in real-time;

[0025] S3. Dynamic adjustment of travel speed

[0026] Compare d_current with the expected grinding depth d_target (pre-set by the worker) to obtain the grinding depth deviation e = d_target - d_current; adopt the PID (Proportional-Integral-Differential) control algorithm to calculate the adjustment amount ΔV of the travel speed: ΔV = Kp * e + Ki * ∫e dt + Kd * de / dt; where Kp, Ki, Kd are PID control parameters; finally, output the target speed to the two drive wheel motors:

[0027] V_target = V_current + ΔV

[0028] When d_current < d_target (grinding too shallow, may encounter protrusions): e > 0, ΔV is positive, reduce the travel speed, which increases the grinding time of the grinding disc in the same area, thereby increasing the actual grinding depth and making it approach the target;

[0029] When d_current > d_target (grinding too deep): e < 0, ΔV is negative, increase the travel speed, which reduces the grinding time, thereby reducing the actual grinding depth;

[0030] When d_current = d_target: maintain the current travel speed;

[0031] S4. Repeat iteration

[0032] Repeat steps S1 to S3 to form a closed-loop control, enabling the grinding machine to automatically maintain the set grinding depth at the optimal travel speed on any terrain.

[0033] As a preferred embodiment of the present invention, the soft measurement model for grinding depth includes an engineering simplified model and a shallow neural network model. The formula for the engineering simplified model is: d=K0*(T / N)*(1 / G)*C+d0, where K0 is the system calibration constant for matching the equipment and the grinding disc, and d0 is the initial zero-point offset when the grinding disc contacts the ground without generating depth. The shallow neural network model uses the grinding motor torque T, rotation speed N, grinding disc mesh number G, and ground material coefficient C as input parameters, and uses calibration experimental grinding depth data as training samples. After training, it outputs high-precision actual grinding depth d_current in real time.

[0034] As a preferred embodiment of the present invention, it further includes a height sensor-assisted control step: 2-3 laser or ultrasonic height sensors are evenly distributed around the grinding disc to collect the vertical distance between the grinding disc and the ground in real time, calculate the direct grinding depth d_direct, use d_direct to perform online parameter correction on the soft measurement model, or switch to the sensor direct control mode when the model fails.

[0035] As a preferred embodiment of the present invention, a Kalman filter algorithm is used to fuse the grinding depth data output by the soft measurement model with the actual grinding depth data measured by the height sensor, and output the optimal grinding depth value for closed-loop control.

[0036] As a preferred embodiment of the present invention, the PID control algorithm can be replaced by a fuzzy control algorithm or a model predictive control algorithm to achieve accurate prediction and control of the travel speed.

[0037] An adaptive control system for the grinding depth of an automatic floor grinder, applicable to the adaptive control method for the grinding depth of the aforementioned automatic floor grinder, includes a grinder body, a sensor module, a control module, and a data storage module. The grinder body includes a frame, a walking mechanism driven independently by two drive wheel motors, a grinding disc and grinding pads driven by a grinding motor. The sensor module includes a current / torque sensor for real-time acquisition of the working current I or output torque T of the grinding motor, and a speed sensor for real-time acquisition of the actual rotational speed N of the grinding motor.

[0038] As a preferred embodiment of the present invention, the data storage module has built-in calibration parameters, system constants, and trained neural network model parameters corresponding to different grinding disc mesh counts and different ground materials; the control module is electrically connected to the grinding machine body, the sensor module, and the data storage module respectively, and is used to collect sensor data, calculate grinding depth, calculate speed adjustment commands, and control the equipment travel speed to realize closed-loop adaptive control of grinding depth.

[0039] As a preferred embodiment of the present invention, it further includes a height sensor for measuring the distance between the grinding disc and the ground. The height sensor is a laser displacement sensor or an ultrasonic displacement sensor, which is evenly distributed on the outer peripheral sidewall of the grinding disc and set vertically toward the ground.

[0040] As a preferred embodiment of the present invention, the control module has a built-in PID control unit, a fuzzy control unit, a data fusion unit, and an online parameter identification unit, which are used to support switching between multiple control algorithms and adaptive optimization of operating conditions.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] Achieving adaptive closed-loop control: For the first time, the load feedback of the grinding motor is introduced into the travel speed control of the automatic floor grinder, enabling the machine to have sensing and adaptive capabilities, fundamentally solving the problem of local unevenness caused by grinding with fixed parameters;

[0043] Improve construction quality and consistency: By using mathematical models to precisely control the grinding depth at each point, the flatness and uniformity of the entire construction surface are far superior to manual or open-loop control methods.

[0044] Reduce hardware costs and system complexity: The main solution utilizes existing motor driver current / torque feedback, eliminating the need for additional expensive high-precision sensors, thus achieving "soft measurement" and having extremely high application value in cost-sensitive commercial products;

[0045] High level of intelligence: It can be seamlessly integrated into existing automatic navigation systems to achieve complete unmanned operation from path planning to process execution, thereby improving construction efficiency and automation level;

[0046] The model is highly scalable: the established mathematical model can be continuously optimized through experimental data, and even machine learning can be introduced to continuously improve the control accuracy as the usage time increases. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the method flow provided in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of the system flow provided in an embodiment of the present invention. Detailed Implementation

[0049] To further understand the invention's content, features, and effects, the following embodiments are provided, and detailed descriptions are given in conjunction with the accompanying drawings.

[0050] The structure of the present invention will now be described in detail with reference to the accompanying drawings.

[0051] refer to Figures 1 to 2As shown in the figure, an adaptive control method for grinding depth of an automatic floor grinder provided by an embodiment of the present invention includes the following steps:

[0052] Core mathematical model (soft measurement model of grinding depth)

[0053] The load torque (T) of the grinding motor during stable operation is related to the grinding power (P) and speed (N) by the following relationship: P = T * N. The grinding power is mainly consumed to overcome the friction between the grinding disc and the ground. This friction is related to the normal force (F_n) of the grinding disc on the ground, the coefficient of friction (μ), the contact area between the grinding disc and the ground, and the relative speed. In automatic grinding machines, the grinding depth (d) directly affects the normal force and the coefficient of friction.

[0054] Therefore, we propose the following functional relationship: T (or I) = f(N,G,C,d), where:

[0055] T: Real-time torque of the grinding motor (or I: real-time current, which is approximately proportional to the torque).

[0056] N: Real-time rotational speed of the grinding motor (rpm)

[0057] G: Current grinding disc mesh size (dimensionless, the larger the value, the finer the grinding disc).

[0058] C: Ground material coefficient (experimental calibration value, such as 1.0 for ordinary concrete and 1.3 for high-hardness emery flooring, etc.).

[0059] d: Real-time grinding depth (mm)

[0060] The formula for calculating the grinding depth (d) is as follows: Through extensive experiments on a specific type of grinding machine, the inverse function of the above function can be fitted, namely: d = F_inv(T, N, G, C)

[0061] A simplified engineering implementation model can be expressed as: d = K0*(T / N)*(1 / G)*C+d0

[0062] K0: System calibration coefficient (constant) related to the equipment and grinding discs.

[0063] d0: Initial offset (zero point when the grinding disc just contacts the ground but has not yet created depth)

[0064] More precisely, considering the nonlinear characteristics of the grinding process, a piecewise function or a machine learning model (such as a neural network) can be used: d=NN_model(T,N,G,C), where NN_model is a pre-trained shallow neural network (input layer: T,N,G,C; output layer: d), and the training data comes from calibration experiments under different working conditions;

[0065] Control method process (combined with the above mathematical model):

[0066] S1. Data collection and preprocessing

[0067] Complete the initialization of system parameters. The control system collects the real-time torque T or working current I and real-time speed N of the grinding motor in real time at a high frequency (such as 100 Hz), reads the current grinding disc mesh number G set by the system and the ground material coefficient C selected through the human-machine interface;

[0068] S2. Real-time grinding depth calculation

[0069] Substitute the collected I (or T), N, G, C into the above formula d = F_inv(T, N, G, C) or the trained neural network model NN_model to calculate the current grinding depth d_current in real time;

[0070] S3. Dynamic adjustment of traveling speed

[0071] Compare d_current with the desired grinding depth d_target (pre-set by the worker) to obtain the grinding depth deviation e = d_target - d_current; adopt the PID (Proportional-Integral-Differential) control algorithm to calculate the adjustment amount ΔV of the traveling speed: ΔV = Kp*e + Ki*∫e dt + Kd*de / dt; where Kp, Ki, Kd are PID control parameters; finally, output the target speed to the two drive wheel motors:

[0072] V_target = V_current + ΔV

[0073] When d_current < d_target (grinding too shallow, may encounter protrusions): e > 0, ΔV is positive, reduce the traveling speed, which increases the grinding time of the grinding disc in the same area, thereby increasing the actual grinding depth and making it approach the target;

[0074] When d_current > d_target (grinding too deep): e < 0, ΔV is negative, increase the traveling speed, which reduces the grinding time, thereby reducing the actual grinding depth;

[0075] When d_current = d_target: Keep the current traveling speed;

[0076] S4. Repeated iteration

[0077] Repeat steps S1 to S3 to form a closed-loop control, so that the grinding machine can automatically maintain the set grinding depth at the optimal traveling speed on any terrain.

[0078] Specifically, the soft measurement model for grinding depth includes an engineering simplified model and a shallow neural network model. The formula for the engineering simplified model is: d=K0*(T / N)*(1 / G)*C+d0, where K0 is the system calibration constant for matching the equipment and the grinding disc, and d0 is the initial zero-point offset when the grinding disc contacts the ground without generating depth. The shallow neural network model uses the grinding motor torque T, rotation speed N, grinding disc mesh number G, and ground material coefficient C as input parameters, and uses calibration experimental grinding depth data as training samples. After training, it outputs high-precision actual grinding depth d_current in real time.

[0079] Specifically, it also includes a height sensor-assisted control step: 2-3 laser or ultrasonic height sensors are evenly distributed around the grinding disc to collect the vertical distance between the grinding disc and the ground in real time, calculate the direct grinding depth d_direct, use d_direct to perform online parameter correction on the soft measurement model, or switch to sensor direct control mode when the model fails.

[0080] Specifically, the Kalman filter algorithm is used to fuse the grinding depth data output by the soft sensor model with the actual grinding depth data measured by the height sensor, and output the optimal grinding depth value for closed-loop control.

[0081] Specifically, the PID control algorithm can be replaced by a fuzzy control algorithm or a model predictive control algorithm to achieve accurate prediction and control of the travel speed.

[0082] An adaptive control system for the grinding depth of an automatic floor grinder, applicable to the adaptive control method for the grinding depth of the aforementioned automatic floor grinder, includes a grinder body, a sensor module, a control module, and a data storage module. The grinder body includes a frame, a walking mechanism driven independently by two drive wheel motors, a grinding disc and grinding pads driven by a grinding motor. The sensor module includes a current / torque sensor for real-time acquisition of the working current I or output torque T of the grinding motor, and a speed sensor for real-time acquisition of the actual rotational speed N of the grinding motor.

[0083] Specifically, the data storage module has built-in calibration parameters, system constants, and trained neural network model parameters corresponding to different grinding disc grit counts and different ground materials; the control module is electrically connected to the grinding machine body, sensor module, and data storage module respectively, and is used to collect sensor data, calculate grinding depth, calculate speed adjustment commands, and control the equipment travel speed to realize closed-loop adaptive control of grinding depth.

[0084] Specifically, it also includes a height sensor for measuring the distance between the grinding disc and the ground. The height sensor is a laser displacement sensor or an ultrasonic displacement sensor, which is evenly distributed on the outer peripheral sidewall of the grinding disc and set vertically toward the ground.

[0085] Specifically, the control module has a built-in PID control unit, fuzzy control unit, data fusion unit, and online parameter identification unit to support switching between multiple control algorithms and adaptive optimization under operating conditions.

[0086] This invention innovatively introduces a grinding motor load feedback mechanism, abandoning the traditional manual control mode. It can be seamlessly integrated with existing equipment's automatic navigation system, achieving unmanned operation of the entire path planning and grinding process. Through multi-parameter fusion modeling, this invention calculates the grinding depth in real time and dynamically adjusts the travel speed for conditions such as localized ground protrusions and uneven material, achieving targeted grinding, eliminating residual protrusions, effectively improving overall ground flatness, and ensuring uniformity and standardization of construction quality over large areas. The core control scheme of this invention relies on the equipment's existing motor sensors to achieve soft measurement of grinding depth, eliminating the need for additional high-precision ranging equipment. With only a small increase in the control algorithm program and extremely low hardware modification costs, this invention is widely compatible with various existing automatic floor grinding machines and has high commercial application value. It simultaneously incorporates a soft-sensor main control scheme and a sensor-based direct measurement auxiliary scheme, with dual modes serving as backups for each other, improving system operational stability. Furthermore, the model can be continuously iterated and optimized using on-site construction data, supporting machine learning algorithm upgrades, and control accuracy can be continuously improved over time. This invention supports switching between multiple algorithms such as PID control, fuzzy control, and model predictive control, and, combined with data fusion and online parameter identification technologies, can adapt to complex and varied floor construction conditions, thus broadening its applicability.

[0087] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0088] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive control of grinding depth in an automatic floor grinder, characterized in that, Includes the following steps: S1. Data Acquisition and Preprocessing: Complete system parameter initialization, input the grinding disc mesh number G and ground material coefficient C corresponding to the current construction condition, and collect the real-time torque T or working current I and real-time speed N of the grinding motor at a preset frequency. S2. Real-time grinding depth calculation: Substitute the collected torque T / current I, rotation speed N, grinding disc mesh number G, and ground material coefficient C into the preset grinding depth soft measurement model to calculate the current actual grinding depth d_current in real time. S3. Dynamic adjustment of travel speed: Obtain the system's preset target grinding depth d_target, calculate the depth deviation value e=d_target -d_current, and calculate the grinding machine travel speed adjustment amount ΔV through the PID control algorithm; then, based on the current travel speed V_current and the speed adjustment amount ΔV, calculate the target travel speed V_target, issue a speed command to adjust the grinding machine's travel speed, and dynamically adapt the grinding time according to the depth deviation so that the actual grinding depth approaches the target grinding depth; S4. Repeated Iteration: Execute steps S1-S3 in a loop to form a closed-loop adaptive control and achieve dynamic and precise control of grinding depth.

2. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 1, characterized in that, The soft measurement model for grinding depth includes an engineering simplified model and a shallow neural network model. The formula for the engineering simplified model is: d=K0*(T / N)*(1 / G)*C+d0, where K0 is the system calibration constant for matching the equipment and the grinding disc, and d0 is the initial zero-point offset of the grinding disc before it contacts the ground and generates depth. The shallow neural network model uses the grinding motor torque T, rotation speed N, grinding disc mesh number G, and ground material coefficient C as input parameters, and uses calibration experimental grinding depth data as training samples. After training, it outputs high-precision actual grinding depth d_current in real time.

3. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 1, characterized in that, It also includes a height sensor-assisted control step: 2-3 laser or ultrasonic height sensors are evenly distributed around the grinding disc to collect the vertical distance between the grinding disc and the ground in real time, calculate the direct grinding depth d_direct, use d_direct to perform online parameter correction on the soft measurement model, or switch to sensor direct control mode when the model fails.

4. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 3, characterized in that, The Kalman filter algorithm is used to fuse the grinding depth data output by the soft measurement model with the actual grinding depth data measured by the height sensor, and output the optimal grinding depth value for closed-loop control.

5. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 1, characterized in that, The PID control algorithm can be replaced by a fuzzy control algorithm or a model predictive control algorithm to achieve accurate prediction and control of travel speed.

6. An adaptive control system for grinding depth of an automatic floor grinder, characterized in that, The adaptive control method for grinding depth of the automatic floor grinder according to any one of claims 1-5 includes a grinder body, a sensor module, a control module, and a data storage module. The grinder body includes a frame, a walking mechanism driven independently by two drive wheel motors, a grinding disc and grinding pads driven by a grinding motor, and the sensor module includes a current / torque sensor for real-time acquisition of the working current I or output torque T of the grinding motor, and a speed sensor for real-time acquisition of the actual rotational speed N of the grinding motor.

7. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 6, characterized in that, The data storage module is used to store equipment calibration parameters, grinding disc parameters, ground material coefficients, and parameters of the trained neural network model. The control module is electrically connected to the grinding machine body, sensor module, and data storage module, respectively, and is used to collect sensor data, calculate grinding depth, calculate speed adjustment commands, and control the equipment travel speed to achieve closed-loop adaptive control of grinding depth.

8. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 6, characterized in that, It also includes a height sensor for measuring the distance between the grinding disc and the ground. The height sensor is a laser displacement sensor or an ultrasonic displacement sensor, which is evenly distributed on the outer peripheral sidewall of the grinding disc and set vertically toward the ground.

9. The adaptive control method for grinding depth of an automatic floor grinder as described in claim 6, characterized in that, The control module incorporates a PID control unit, a fuzzy control unit, a data fusion unit, and an online parameter identification unit to support switching between multiple control algorithms and adaptive optimization under operating conditions.

Images