A sandblasting process self-adaptive control system based on roughness feedback

Abstract

The application discloses a sand blasting process self-adaptive control system based on roughness feedback, belongs to the technical field of surface treatment equipment and sand blasting process automatic control, and a surface partition modeling module is used for dividing the surface of a workpiece to be blasted into a plurality of surface units, and establishing target roughness, measured roughness and historical sand blasting state for each surface unit; an online roughness detection module is arranged behind a sand blasting action area of a nozzle, is used for acquiring the measured roughness of the surface unit after sand blasting, and sends the measured roughness to a self-adaptive controller; and a sand blasting energy distribution module is used for generating target sand blasting energy of each surface unit according to the difference between the target roughness and the measured roughness of each surface unit and the historical sand blasting state of the corresponding surface unit; and the application changes the sand blasting control object from the whole workpiece to the independently managed surface unit, thereby being capable of performing differential control according to the surface state difference of different regions and reducing local under-blasting and over-blasting.

Description

Technical Field

[0001] This invention relates to the field of surface treatment equipment and automatic control technology for sandblasting processes, and particularly to an adaptive control system for sandblasting processes based on roughness feedback. Background Technology

[0002] Sandblasting is a surface treatment process that uses high-speed abrasive particles to impact the surface of a workpiece to remove oxide layers, rust layers, contaminant layers, or adjust surface roughness. The surface roughness after sandblasting affects subsequent coating adhesion, surface bonding, sealing fit, and appearance consistency.

[0003] Existing sandblasting equipment typically treats workpiece surfaces according to preset sandblasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed. In actual processing, due to factors such as inconsistent initial surface conditions, differences in local morphology, variations in abrasive particle size and wear state, nozzle wear, and uneven distribution of the sandblasting jet, even with the same sandblasting parameters, the roughness response in different areas may vary significantly.

[0004] For large steel structural components, complex curved surfaces, or workpieces with significant local obstructions, fixing the sandblasting parameters can easily lead to some areas failing to achieve the target roughness range, while other areas may experience roughness overshoot, pitting, or substrate damage due to excessive energy input. Relying on manual inspection and rework correction will increase processing cycle time and abrasive consumption, and reduce the consistency of surface treatment between different areas.

[0005] Furthermore, there is usually a spatial distance between the blasting point and the roughness detection point. When the nozzle scans continuously, there is a time or spatial delay in the online roughness detection results relative to the actual blasting area. Existing control methods struggle to accurately register the detected roughness to the corresponding blasted surface area, resulting in delayed feedback adjustment and further increasing the risk of over-blasting or under-blasting.

[0006] Meanwhile, existing systems adjust based solely on the deviation between the measured roughness and the target value, lacking the ability to predict roughness trends and thus prone to overshoot or oscillations during adjustment. Furthermore, the determination of blasting parameter combinations often relies on experience or a single objective, failing to comprehensively consider the balance between processing efficiency and abrasive consumption.

[0007] Therefore, it is necessary to provide a sandblasting process control system that can adaptively allocate sandblasting energy to different surface regions based on surface roughness feedback, historical status, trend prediction, and multi-objective optimization, in order to improve the consistency of roughness control and reduce the risk of local over-blasting. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide an adaptive control system for the sandblasting process based on roughness feedback. It aims to solve the problems of insufficient roughness or roughness overshoot caused by the difference in the initial state of the surface, the nonlinear relationship between sandblasting energy and roughness response, and the time and space delay of online detection feedback when the workpiece surface is treated with fixed sandblasting parameters in the prior art.

[0009] To achieve the above-mentioned objectives, the first aspect of the present invention proposes an adaptive control system for a sandblasting process based on roughness feedback, comprising a sandblasting execution module and an adaptive controller. The sandblasting execution module includes a nozzle, a pressure regulating unit, an abrasive flow regulating unit, a nozzle distance regulating unit, a nozzle angle regulating unit, and a scanning motion unit. It also includes a surface partitioning modeling module, an online roughness detection module, a sandblasting energy distribution module, a roughness trend prediction module, and a sandblasting strategy optimization module. The surface partitioning modeling module is used to divide the surface of the workpiece to be sandblasted into multiple surface units, and to establish the target roughness, measured roughness and historical sandblasting status for each surface unit. The online roughness detection module is located after the sandblasting area of ​​the nozzle, and is used to obtain the measured roughness of the surface unit after sandblasting, and send the measured roughness to the adaptive controller. The sandblasting energy distribution module is used to generate the target sandblasting energy for each surface unit based on the difference between the target roughness and the measured roughness of each surface unit and the historical sandblasting state of the corresponding surface unit. The roughness trend prediction module is used to generate the predicted roughness for the next moment or the next scanning cycle based on the historical measured roughness sequence of each surface unit using a time series prediction algorithm, and send the predicted roughness to the sandblasting energy distribution module for advance correction of the target sandblasting energy. The sandblasting strategy optimization module is used to generate an optimal combination of parameters for sandblasting pressure, abrasive flow rate, nozzle distance, spray angle and scanning speed by using the roughness of each surface unit as a constraint, the shortest total sandblasting time and the minimum total abrasive consumption as optimization objectives, and adopting a multi-objective optimization algorithm. The optimal parameter combination is then sent to the adaptive controller as a reference benchmark for the collaborative control command. It also includes a sandblasting actuator dynamic response compensation module, which is set between the adaptive controller and the sandblasting actuator module. It is used to identify, predict and compensate for the response delay and dynamic response characteristics of each adjustment unit, and generate a dynamically compensated collaborative control command. The adaptive controller generates coordinated control commands for blasting pressure, abrasive flow rate, nozzle distance, injection angle, and scanning speed based on the target blasting energy, the predicted roughness, and the preferred parameter combination, and sends the coordinated control commands to the dynamic response compensation module of the blasting actuator. After the dynamic response compensation module of the blasting actuator performs dynamic response compensation on the coordinated control commands, it controls the blasting actuator to perform adaptive blasting processing on the corresponding surface unit.

[0010] Optionally, the dynamic response compensation module of the sandblasting actuator includes an actuator identification unit, a dynamic response prediction unit, a look-ahead command pre-compensation unit, and a multi-actuator synchronous registration unit; The actuator identification unit is used to apply excitation signals to each regulating unit during the system startup or maintenance phase, collect the actual response data of each regulating unit, identify the transfer function model or state space model of each regulating unit, and the identification parameters include at least two of the following: delay time, rise time, overshoot, steady-state error, and response bandwidth of each regulating unit. The dynamic response prediction unit is used to predict the actual output trajectory of each control unit in the future multiple control cycles based on the current state of each control unit and the cooperative control command sequence, and send the predicted actual output to the adaptive controller or the look-ahead command pre-compensation unit. The forward-looking command pre-compensation unit is used to perform advance correction or inverse model compensation on the coordinated control command based on the dynamic response prediction results, so that the actual output of each adjustment unit responds in advance to offset the inherent delay. The multi-actuator synchronization registration unit is used to calculate the difference between the response delays of each control unit, and with the control unit with the slowest response as the benchmark, adds a delay queue to the control commands of other control units, so that each control unit can achieve synchronous response at the spatial position of the sandblasting point.

[0011] Optionally, the excitation signal used by the actuator identification unit includes a step signal, a sinusoidal sweep signal, or a pseudo-random binary sequence signal; the transfer function model is a first-order inertial plus pure delay model or a second-order oscillation plus pure delay model; and the state space model is a linear state space model or a Hammerstein model with input nonlinearity.

[0012] Optionally, the look-ahead command pre-compensation unit adopts zero-pole cancellation compensation, lead-lag correction compensation, or model predictive control compensation based on the inverse model; when the target sandblasting energy undergoes a step change, the look-ahead command pre-compensation unit issues a pressure increase or decrease command in advance according to the response delay of the pressure adjustment unit, and synchronously adjusts the time sequence of the scanning speed command according to the acceleration and deceleration characteristics of the scanning motion unit.

[0013] Optionally, the multi-actuator synchronous registration unit uses the scanning motion unit as the reference actuator, calculates the response delay difference between the pressure adjustment unit, abrasive flow adjustment unit, nozzle distance adjustment unit and nozzle angle adjustment unit and the scanning motion unit, and adds the corresponding delay amount to the control command of each adjustment unit, so that the actual output of all adjustment units reaches the target state at the moment when the sandblasting point reaches the target surface unit.

[0014] Optionally, the dynamic response compensation module of the sandblasting actuator further includes an actuator state monitoring subunit, which is used to monitor the actual response output of each adjustment unit in real time and compare the actual response output with the predicted output of the dynamic response prediction unit; when the deviation between the actual response and the predicted response exceeds a set threshold, the actuator state monitoring subunit triggers the actuator identification unit to re-identify the model in order to update the dynamic model parameters.

[0015] Optionally, the dynamic response prediction unit is also used to predict the overshoot and oscillation trend of each adjustment unit under rapid energy switching conditions; when it is predicted that the overshoot of the transition process will cause the sandblasting energy to exceed the allowable deviation range of the target sandblasting energy, the look-ahead command pre-compensation unit adds a smoothing constraint to the control command to limit the sandblasting energy change rate to within the safe change rate range.

[0016] Optionally, the historical sandblasting status includes at least two of the following for the corresponding surface unit: cumulative sandblasting time, cumulative sandblasting energy, previous sandblasting pressure, previous abrasive flow rate, previous nozzle distance, previous spray angle, previous scanning speed, and previous measured roughness.

[0017] Optionally, it also includes a surface unit memory module, which is used to store the spatial position, sandblasting trajectory, sandblasting execution parameters, measured roughness and historical sandblasting status of each surface unit, and to match the measured roughness obtained by the online roughness detection module with the corresponding surface unit; The surface unit memory module includes a delay registration unit, which is used to register the spatial or temporal delay between sandblasting and roughness detection based on the nozzle position, nozzle moving speed, sandblasting point position, and roughness detection point position.

[0018] Optionally, it also includes an anomaly protection module, which is used to control the sandblasting execution module to perform at least one of the following protective actions when the measured roughness exceeds the target roughness upper limit, the roughness detection signal is abnormal, the nozzle distance is abnormal, the abrasive flow rate is abnormal, or the nozzle wear is abnormal: reducing the sandblasting pressure, reducing the abrasive flow rate, increasing the nozzle distance, increasing the scanning speed, or pausing sandblasting.

[0019] The beneficial effects of this invention are: 1. The adaptive control system for sandblasting based on roughness feedback of the present invention divides the workpiece surface into multiple surface units through a surface partitioning modeling module, and establishes target roughness, measured roughness, and historical sandblasting state for each surface unit. This transforms the sandblasting control object from the entire workpiece to independently manageable surface units, enabling differentiated control based on surface state differences in different areas and reducing local under-blasting and over-blasting. The measured roughness after sandblasting is obtained through an online roughness detection module, and a target sandblasting energy is generated based on the roughness difference and historical sandblasting state through a sandblasting energy distribution module. This transforms the sandblasting control objective from adjusting single equipment parameters to energy distribution control based on surface state, which is beneficial for converging the roughness of different surface units towards the target range.

[0020] 2. The roughness feedback-based adaptive control system for sandblasting process of the present invention converts the target sandblasting energy into coordinated control commands of sandblasting pressure, abrasive flow rate, nozzle distance, jet angle and scanning speed through an adaptive controller, thereby avoiding excessive or insufficient response caused by relying on adjustment of a single parameter, thus improving the control stability of the sandblasting process.

[0021] 3. The adaptive control system for sandblasting process based on roughness feedback of the present invention matches the online roughness detection results with the corresponding sandblasted surface units through the surface unit memory module and the delay registration unit, which can reduce the impact of spatial or temporal delay between the sandblasting point and the detection point on the accuracy of feedback control.

[0022] 4. The adaptive control system for sandblasting process based on roughness feedback of the present invention performs protective actions when the roughness exceeds the limit, the detection signal is abnormal, the nozzle distance is abnormal, the abrasive flow rate is abnormal, or the nozzle wear is abnormal through the abnormal protection module. This can reduce the risk of substrate damage during continuous sandblasting and improve the safety of system operation.

[0023] 5. By adding a roughness trend prediction module, this invention uses historical roughness sequences to predict future trends, enabling the control system to have foresight and adjust energy input in advance before the deviation expands, effectively suppressing overshoot and oscillation, and improving the smoothness of roughness convergence.

[0024] 6. This invention adds a sandblasting strategy optimization module, with roughness compliance as a constraint and shortest time and least wear as multiple objectives. It uses an evolutionary algorithm to solve for the optimal parameter combination, overcoming the blindness of the traditional trial-and-error method. While ensuring quality, it significantly improves processing efficiency and reduces abrasive costs.

[0025] 7. The adaptive control system for sandblasting process based on roughness feedback of the present invention identifies, predicts, and compensates for the dynamic response characteristics of each regulating unit through a dynamic response compensation module for sandblasting actuators. This eliminates the spatiotemporal mismatch of sandblasting parameters caused by differences in physical response delays among pressure regulating units, abrasive flow regulating units, and scanning motion units. Through the advance correction of the look-ahead command pre-compensation unit, slower-responding actuators are activated earlier. Through the delay queue registration of the multi-actuator synchronization registration unit, each actuator reaches the target state synchronously at the sandblasting point, effectively suppressing parameter mismatch during the rapid energy switching transition of surface units and improving the response accuracy and stability of the roughness closed-loop control.

[0026] 8. The adaptive control system for sandblasting process based on roughness feedback of the present invention compares the actual response with the predicted response in real time through the actuator state monitoring subunit, and triggers the actuator identification unit to re-identify the model when the deviation exceeds the threshold, so that the dynamic compensation model can adaptively track the response characteristic drift caused by factors such as wear of the regulating valve, changes in hydraulic oil temperature, and aging of the actuator.

[0027] 9. The adaptive control system for sandblasting process based on roughness feedback of the present invention predicts the overshoot and oscillation trend under rapid energy switching conditions through a dynamic response prediction unit, and adds smoothing constraints to the control command through a look-ahead command pre-compensation unit, thereby limiting the rate of change of sandblasting energy and avoiding instantaneous over-blasting or under-blasting caused by actuator response overshoot, thus reducing the risk of overshoot and oscillation in the roughness control process. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the system architecture of the adaptive control system for sandblasting process based on roughness feedback according to the present invention. Figure 2 This is a schematic diagram of the sandblasting execution module of the sandblasting process adaptive control system based on roughness feedback according to the present invention; Figure 3 This is a schematic diagram of the online roughness detection module of the sandblasting process adaptive control system based on roughness feedback of the present invention; Figure 4 This is a schematic diagram of the abrasive state monitoring module of the adaptive control system for sandblasting process based on roughness feedback according to the present invention. Figure 5 This is a schematic diagram of the workpiece surface unit division and sandblasting trajectory of the adaptive control system for sandblasting process based on roughness feedback according to the present invention. Figure 6 This is a schematic diagram of the sandblasting energy distribution module of the sandblasting process adaptive control system based on roughness feedback of the present invention. Figure 7This is a schematic diagram of the dynamic response compensation module of the sandblasting actuator in the adaptive control system for sandblasting process based on roughness feedback of the present invention.

[0029] Explanation of reference numerals in the attached figures: 10. Workpiece; 11. Surface unit; 12. Sandblasting trajectory; 13. Sandblasting point; 14. Roughness detection point; 20. Sandblasting execution module; 21. Nozzle; 22. Pressure adjustment unit; 23. Abrasive flow rate adjustment unit; 24. Nozzle distance adjustment unit; 25. Nozzle angle adjustment unit; 26. Scanning motion unit; 30. Online roughness detection module; 31. Laser detection unit; 32. Protective window; 33. Air curtain isolation unit; 40. Sandblasting energy distribution module; 41. Roughness difference calculation unit; 42. Historical state correction unit; 43. Target sandblasting energy generation unit; 50. Roughness trend prediction module; 60. Abrasive condition monitoring module; 61. Abrasive flow sensor; 62. Nozzle inlet pressure sensor; 63. Nozzle outlet acoustic sensor; 64. Dust concentration sensor; 70. Adaptive controller; 80. Anomaly Protection Module; 90. Surface Partition Modeling Module; 100. Surface unit memory module; 101. Delayed registration unit; 110. Sandblasting strategy optimization module; 120. Sandblasting actuator dynamic response compensation module; 121. Actuator identification unit; 122. Dynamic response prediction unit; 123. Look-ahead command pre-compensation unit; 124. Multi-actuator synchronization registration unit; 125. Actuator status monitoring subunit.

[0030] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0031] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0032] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection, a direct connection, or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0033] Reference Figures 1-7 This embodiment provides an adaptive control system for a sandblasting process based on roughness feedback, including a sandblasting execution module 20, an adaptive controller 70, a surface partitioning modeling module 90, an online roughness detection module 30, a sandblasting energy distribution module 40, a roughness trend prediction module 50, and a sandblasting strategy optimization module 110.

[0034] The sandblasting execution module 20 is used to perform sandblasting on the surface of the workpiece 10. The sandblasting execution module 20 includes a nozzle 21, a pressure adjustment unit 22, an abrasive flow rate adjustment unit 23, a nozzle distance adjustment unit 24, a nozzle angle adjustment unit 25, and a scanning motion unit 26.

[0035] Nozzle 21 is used to spray a high-speed gas-solid two-phase flow carrying abrasive particles onto the surface of workpiece 10. Nozzle 21 can be a wear-resistant ceramic nozzle, boron carbide nozzle, tungsten carbide nozzle, or other wear-resistant nozzles suitable for sandblasting. Pressure regulating unit 22 may include a proportional pressure reducing valve, an electro-proportional valve, or a servo pressure regulating valve for regulating the nozzle inlet pressure. Abrasive flow regulating unit 23 may include an abrasive supply valve, a screw feed mechanism, a vibrating feed mechanism, or a negative pressure suction regulating mechanism for regulating the amount of abrasive entering nozzle 21 per unit time. Nozzle distance regulating unit 24 may include a linear slide, an electric cylinder, a pneumatic slide, or a robot end effector for regulating the distance between the nozzle 21 outlet and the surface of workpiece 10. Nozzle angle regulating unit 25 may include a rotary joint, a swing mechanism, or a multi-axis robot posture regulating mechanism for regulating the spray angle of nozzle 21 relative to the surface of workpiece 10. Scanning motion unit 26 is used to move nozzle 21 along sandblasting trajectory 12. Scanning motion unit 26 can be a gantry motion platform, a robotic arm, a CNC slide, or a dedicated scanning mechanism.

[0036] like Figure 1As shown, the surface partitioning modeling module 90 is used to divide the surface of the workpiece 10 to be sandblasted into multiple surface units 11. Each surface unit 11 can be a rectangular mesh unit, a curved mesh unit, a point cloud neighborhood unit, or a strip unit divided along the sandblasting trajectory 12. For planar workpieces, the surface units 11 can be divided according to a two-dimensional coordinate mesh; for curved workpieces, the surface units 11 can be divided according to a three-dimensional point cloud, surface parameter coordinates, or robot trajectory coordinates. Each surface unit 11 has a corresponding spatial position, target roughness, measured roughness, and historical sandblasting state.

[0037] In an optional implementation, the surface partitioning modeling module 90 can also record the initial roughness of each surface unit 11. The initial roughness is used to determine the initial sandblasting energy level before the first sandblasting. In subsequent closed-loop control, the system uses the measured roughness and historical sandblasting status obtained by the online roughness detection module 30 as the main feedback basis.

[0038] Target roughness can be expressed as a target value or a target range. For example, for pretreatment of large steel structural parts before painting, the target roughness can be set to Ra 6.3 μm to Ra 12.5 μm; for precision surface treatment of molds or aerospace parts, the target roughness can be set to Ra 2.5 μm to Ra 6.3 μm. The allowable range of target roughness can be determined according to the workpiece material, coating system, or subsequent assembly requirements.

[0039] The online roughness detection module 30 is positioned after the sandblasting area of ​​the nozzle 21. "After" means that in the direction the nozzle 21 moves along the sandblasting trajectory 12, the detection position of the online roughness detection module 30 is located behind the sandblasting point 13, allowing the online roughness detection module 30 to detect the sandblasted surface unit 11 after the nozzle 21 completes its sandblasting action. The online roughness detection module 30 may include a laser detection unit 31, a protective window 32, and an air curtain isolation unit 33. The laser detection unit 31 is used to acquire height data, scattering data, or contour data of the workpiece 10 surface, and calculate the measured roughness of the surface unit 11 based on the data. The protective window 32 is positioned at the front end of the optical path of the laser detection unit 31 to block abrasive particles and dust. The air curtain isolation unit 33 forms an airflow barrier outside the protective window 32 to reduce the impact of dust adhesion on the detection results.

[0040] The sandblasting energy distribution module 40 is used to generate the target sandblasting energy for each surface unit 11 based on the difference between the target roughness and the measured roughness of each surface unit 11 and the historical sandblasting state of the corresponding surface unit 11. The target sandblasting energy is a control quantity used to characterize the strength of the sandblasting action and is not limited to a single energy value in a strictly physical sense. The target sandblasting energy can establish a correspondence with sandblasting pressure, abrasive flow rate, nozzle distance, jet angle, scanning speed, and action time through calibration curves, parameter mapping tables, empirical control models, or closed-loop control algorithms.

[0041] In one possible implementation, the target blasting energy can be represented as an energy level. Energy levels include low, medium, and high energy levels. The adaptive controller 70 calls upon the corresponding set of parameters—blasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed—based on the energy level. In another possible implementation, the target blasting energy can be represented as a continuous control quantity, and the adaptive controller 70 performs limited and coordinated adjustments to the blasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed based on this continuous control quantity.

[0042] The roughness trend prediction module 50 is used to generate predicted roughness for the next moment or the next scanning cycle based on the historical measured roughness sequence of each surface unit 11 using a time series prediction algorithm. In one embodiment, an exponential smoothing method is used to assign decreasing weights to the most recent N measured roughness values, and the predicted value is used as the expectation of the roughness in the next cycle. In another embodiment, an ARIMA model is used, and the model order is determined through autocorrelation and partial autocorrelation analysis for short-term prediction. In yet another embodiment, an LSTM neural network is used, and the prediction model is trained using a large amount of historical data to adapt to the nonlinear roughness variation law. The predicted roughness and its confidence interval are sent to the sandblasting energy allocation module 40 to pre-correct the target sandblasting energy. Specifically, when the predicted roughness will exceed the upper limit of the target roughness in the next cycle, the sandblasting energy allocation module 40 reduces the target sandblasting energy in advance to avoid overshoot; when the predicted roughness is still below the lower limit of the target roughness in the next cycle, the sandblasting energy allocation module 40 increases the target sandblasting energy in advance to accelerate convergence.

[0043] The sandblasting strategy optimization module 110 is used to generate optimal parameter combinations for sandblasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed, with the roughness of each surface unit 11 meeting the target as a constraint and the shortest total sandblasting time and the minimum total abrasive consumption as optimization objectives. The optimization variables are sandblasting pressure P, abrasive flow rate Q, nozzle distance D, jet angle θ, and scanning speed v. The constraints include: the predicted roughness Ra_pred of each surface unit is within the target range [Ra_low, Ra_high]; the sandblasting pressure P does not exceed the maximum allowable pressure P_max of the equipment; the abrasive flow rate Q is within the supply capacity [Q_min, Q_max]; the nozzle distance D is within the safe distance range; the jet angle θ is within the mechanical limit; and the scanning speed v is within the capability of the motion system. The objective function is to minimize the total blasting time T_total = Σ(L_i / v_i) (where L_i is the trajectory length of the i-th surface element and v_i is the scanning speed of that element), and minimize the total abrasive consumption M_total = Σ(Q_i) (where Q_i is the abrasive flow rate and t_i is the action time). The Pareto optimal solution set is solved using a non-dominated sorting genetic algorithm (NSGA-II) or a multi-objective particle swarm optimization algorithm (MOPSO). The user or system selects a compromise solution as the preferred parameter combination based on the current operating conditions (such as production cycle requirements and abrasive cost weights), and sends it to the adaptive controller 70 as a reference for cooperative control commands.

[0044] In this embodiment, the system further includes a sandblasting actuator dynamic response compensation module 120. The sandblasting actuator dynamic response compensation module 120 is disposed between the adaptive controller 70 and the sandblasting execution module 20. The cooperative control commands generated by the adaptive controller 70 are first sent to the sandblasting actuator dynamic response compensation module 120, and after dynamic response compensation, are output to each adjustment unit of the sandblasting execution module 20.

[0045] Because the pressure regulating unit 22, abrasive flow regulating unit 23, nozzle distance regulating unit 24, nozzle angle regulating unit 25, and scanning motion unit 26 have different physical response characteristics—for example, the response time of the pressure regulating valve and the acceleration / deceleration time of the scanning motion platform are usually mismatched—when the target sandblasting energy rapidly switches between adjacent surface units, the actual output of each actuator may not simultaneously reach the target state. This results in a mismatch in the sandblasting parameter combination during the transition process, affecting the accuracy and stability of roughness control. The sandblasting actuator dynamic response compensation module 120 eliminates the spatiotemporal mismatch of sandblasting parameters caused by actuator response differences by identifying, predicting, and compensating for the dynamic response characteristics of each regulating unit.

[0046] like Figure 7As shown, the dynamic response compensation module 120 for sandblasting actuators includes an actuator identification unit 121, a dynamic response prediction unit 122, a look-ahead command pre-compensation unit 123, a multi-actuator synchronization registration unit 124, and an actuator status monitoring subunit 125.

[0047] The actuator identification unit 121 is used to apply excitation signals to each regulating unit during the system startup or maintenance phase, collect the actual response data of each regulating unit, and identify the transfer function model or state-space model of each regulating unit. The excitation signals include step signals, sinusoidal sweep signals, or pseudo-random binary sequence signals. Taking the pressure regulating unit 22 as an example, the actuator identification unit 121 applies a step voltage signal to the pressure regulating unit 22, and simultaneously collects the actual response curve of the nozzle inlet pressure through a pressure sensor. A first-order inertial plus pure delay model is obtained by fitting the curve using the least squares identification method. G(s) = K·e^(-τs) / (T·s + 1); Where K is the gain, τ is the pure delay time, and T is the inertial time constant. For the scanning motion unit 26, due to its acceleration and deceleration characteristics, a second-order oscillation plus pure delay model can be used for identification. The identification parameters include the delay time τ, rise time, overshoot, steady-state error, and response bandwidth of each adjustment unit. For the abrasive flow adjustment unit 23, due to the inertia of the abrasive particles in the pipeline, its transfer function is usually expressed as a first-order inertial plus delay model with integral characteristics. For the nozzle distance adjustment unit 24, its response characteristics can be identified as a nonlinear model with hysteresis and dead zone. The actuator identification unit 121 sends the dynamic model parameters identified by each adjustment unit to the dynamic response prediction unit 122.

[0048] The dynamic response prediction unit 122 is used to predict the actual output trajectory of each regulating unit in the future multiple control cycles based on the current state of each regulating unit and the cooperative control command sequence output by the adaptive controller 70, using the identified dynamic model. For example, when the adaptive controller 70 issues a command to increase the pressure from 0.4MPa to 0.7MPa in a step at the surface unit switching boundary, the dynamic response prediction unit 122 calculates the rise curve of the actual pressure value in subsequent control cycles based on the identified model of the pressure regulating unit 22, including the delay time, rise slope and overshoot, and sends the predicted actual output to the look-ahead command pre-compensation unit 123.

[0049] The dynamic response prediction unit 122 is also used to predict the overshoot and oscillation trend of each regulating unit during the transition process under rapid energy switching conditions. For example, when the target blasting energy of an adjacent surface unit switches from a high energy level to a low energy level, the pressure regulating unit 22 may experience pressure overshoot due to valve response lag, the scanning motion unit 26 may be unable to quickly reach the target scanning speed due to acceleration limitations, and the abrasive flow regulating unit 23 may experience feeding delay due to the inertia of the feeding mechanism. The dynamic response prediction unit 122 integrates the predicted output curves of each regulating unit to assess the degree to which the actual blasting energy deviates from the target blasting energy during the transition process, and sends the prediction results to the look-ahead command pre-compensation unit 123 and the multi-actuator synchronization registration unit 124.

[0050] The look-ahead command pre-compensation unit 123 is used to perform advance correction or inverse model compensation on the cooperative control command based on the dynamic response prediction results. In one embodiment, the look-ahead command pre-compensation unit 123 adopts a zero-pole cancellation compensation strategy based on the inverse model: for a first-order inertial plus pure delay model, an advance compensation link is connected in series in the controller to speed up the response speed of the compensated system and offset the inherent delay. In another embodiment, the look-ahead command pre-compensation unit 123 adopts a model predictive control compensation strategy, with the goal of minimizing the deviation between the predicted output and the target output, solving for the optimal control input sequence, and outputting the first control quantity in the sequence to the actuator.

[0051] In the scenario where the pressure jumps from 0.4 MPa to 0.7 MPa, if the delay time of the pressure adjustment unit 22 is 80 ms, and the acceleration / deceleration time of the scanning motion unit 26 from the current speed to the target speed is 120 ms, there is a 40 ms time difference between the two. The look-ahead command pre-compensation unit 123, based on the acceleration / deceleration curve of the scanning motion unit 26, sends a pressure increase command to the pressure adjustment unit 22 80 ms in advance, aligning the actual pressure reaching 0.7 MPa with the scanning speed reaching the target value, thereby avoiding mismatches such as "high pressure, low speed" or "low pressure, high speed" during the transition.

[0052] When it is predicted that the overshoot of the transient process will cause the actual sandblasting energy to exceed the allowable deviation range of the target sandblasting energy, the look-ahead command pre-compensation unit 123 adds smoothing constraints to the control command, such as adding S-shaped velocity planning or first-order filtering in the command sequence, so as to limit the sandblasting energy change rate within the safe change rate range, thereby reducing the negative impact of overshoot and oscillation on roughness control.

[0053] The multi-actuator synchronization registration unit 124 calculates the difference in response delay between each regulating unit and adds a delay queue to the control commands of the other regulating units, using the slowest-responding regulating unit as a reference, so that each regulating unit achieves synchronous response at the spatial position of the sandblasting point. Specifically, the multi-actuator synchronization registration unit 124 uses the scanning motion unit 26 as the reference actuator (because its acceleration and deceleration time is usually the longest) and calculates the response delay difference Δτ_i of the pressure regulating unit 22, abrasive flow regulating unit 23, nozzle distance regulating unit 24, and nozzle angle regulating unit 25 relative to the scanning motion unit 26. For regulating units with a faster response than the reference actuator, the multi-actuator synchronization registration unit 124 adds a delay amount Δτ_i to the beginning of their control commands; for regulating units with a response slower than or equal to the reference actuator, no additional delay is added. After delay registration, when the sandblasting point reaches the target surface unit, the actual output of all regulating units simultaneously reaches the target state.

[0054] In one embodiment, the multi-actuator synchronization registration unit 124 also dynamically adjusts the instruction delay queue of each adjustment unit. When the scanning speed changes, the dwell time of the sandblasting point between different surface units changes. The multi-actuator synchronization registration unit 124 recalculates the delay registration amount of each adjustment unit according to the real-time scanning speed and performs interpolation adjustment in the control instruction sequence.

[0055] The actuator condition monitoring subunit 125 monitors the actual response output of each regulating unit in real time (such as pressure sensor readings, abrasive flow sensor readings, encoder position feedback, angle encoder readings, etc.) and compares the actual response output with the predicted output of the dynamic response prediction unit 122 in real time. When the deviation between the actual response and the predicted response exceeds a set threshold (e.g., steady-state error exceeds ±5% or rise time deviation exceeds ±10%), the actuator condition monitoring subunit 125 determines that the actuator dynamic characteristics have changed significantly and triggers the actuator identification unit 121 to re-identify the model to update the dynamic model parameters. This adaptive update mechanism can compensate for the drift in response characteristics caused by factors such as control valve wear, hydraulic oil temperature changes, pneumatic system pressure fluctuations, actuator mechanical wear and aging, so that the dynamic compensation model always remains consistent with the actual actuator state.

[0056] The adaptive controller 70 is communicatively connected to the surface partitioning modeling module 90, the online roughness detection module 30, the sandblasting energy distribution module 40, the roughness trend prediction module 50, the sandblasting strategy optimization module 110, and the sandblasting execution module 20. The adaptive controller 70 can be an industrial control computer, a programmable logic controller, a motion controller, an embedded controller, or an edge computing controller. After receiving the target sandblasting energy, predicted roughness, and optimized parameter combinations for each surface unit 11, the adaptive controller 70 generates coordinated control commands for sandblasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed, and controls the sandblasting execution module 20 to perform adaptive sandblasting processing on the corresponding surface unit 11.

[0057] The working principle of this embodiment is as follows: First, the surface partitioning modeling module 90 divides the surface of the workpiece 10 into multiple surface units 11 based on the geometric model, scanning path, or detection coordinates of the workpiece 10, and sets a target roughness range for each surface unit 11.

[0058] Secondly, the sandblasting strategy optimization module 110 solves a multi-objective optimization problem offline or online based on the overall size of the workpiece, the target roughness requirements, and the equipment capabilities, and obtains a set of preferred parameter combinations as a global benchmark.

[0059] Subsequently, the online roughness detection module 30 detects the surface unit 11 after sandblasting and obtains the measured roughness. The measured roughness is sent to the adaptive controller 70 and further input to the sandblasting energy distribution module 40 and the roughness trend prediction module 50.

[0060] The roughness trend prediction module 50 updates the historical sequence of the surface unit, outputs the predicted roughness, and determines whether the energy needs to be adjusted in advance in the next cycle.

[0061] Then, the sandblasting energy distribution module 40 calculates the difference between the target roughness and the measured roughness, and generates the target sandblasting energy by combining the historical sandblasting state of the surface unit 11 and the predicted roughness. When the measured roughness is lower than the lower limit of the target roughness, it indicates that the sandblasting effect of the surface unit 11 is insufficient, and the sandblasting energy distribution module 40 increases the target sandblasting energy for subsequent sandblasting treatments of the surface unit 11. When the measured roughness is within the allowable range of the target roughness, it indicates that the surface unit 11 has reached the target treatment state, and the sandblasting energy distribution module 40 reduces the target sandblasting energy of the surface unit 11, causing it to enter a state of sustained sandblasting, low-energy trimming, or skipping the treatment. When the measured roughness is higher than the upper limit of the target roughness, it indicates that the surface unit 11 has a risk of roughness overshoot, and the sandblasting energy distribution module 40 generates a control result to stop sandblasting or reduce the sandblasting energy input. At the same time, if the predicted roughness indicates that the next cycle will exceed the limit, the target sandblasting energy is reduced in advance even if the current measured value is still within the range, achieving proactive control.

[0062] Finally, the adaptive controller 70 integrates the target blasting energy, the predicted roughness correction, and the optimal parameter combination provided by the blasting strategy optimization module to generate coordinated control commands for blasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed. For surface units 11 that require increased blasting energy, one or more of the following methods can be used: increasing blasting pressure, increasing abrasive flow rate, decreasing nozzle distance, adjusting jet angle, or decreasing scanning speed. For surface units 11 that require reduced blasting energy, one or more of the following methods can be used: decreasing blasting pressure, decreasing abrasive flow rate, increasing nozzle distance, adjusting jet angle, or increasing scanning speed.

[0063] Then, the dynamic response compensation module 120 of the sandblasting actuator predicts the actual output response under the current control command based on the identification model of each regulating unit. The look-ahead command pre-compensation unit 123 performs advance correction on the commands of each regulating unit: for regulating units with slow response (such as pressure regulating unit 22), the command is issued in advance so that its actual output reaches the target value when the sandblasting point is reached; the multi-actuator synchronization registration unit 124 uses the slowest responding regulating unit (usually the scanning motion unit 26) as the reference and adds a delay queue to other regulating units to ensure that each regulating unit reaches the target state synchronously in the spatial position of the sandblasting point. The dynamically compensated cooperative control command is output to the sandblasting actuator module 20.

[0064] During the sandblasting process, the actuator status monitoring subunit 125 continuously monitors the actual output of each adjustment unit and compares it in real time with the predicted value of the dynamic response prediction unit 122. When the deviation exceeds the threshold, the actuator identification unit 121 is triggered to re-identify the model parameters to ensure that the compensation model always adapts to the actual response characteristics of the current actuator.

[0065] This enables closed-loop control of the sandblasting process, driven by surface roughness feedback and guaranteed by actuator dynamic response compensation.

[0066] like Figure 6 As shown, the sandblasting energy distribution module 40 includes a roughness difference calculation unit 41, a historical state correction unit 42, and a target sandblasting energy generation unit 43.

[0067] The roughness difference calculation unit 41 is used to calculate the difference between the target roughness and the measured roughness of each surface unit 11. When the target roughness is expressed as a range, the roughness difference can be determined based on the deviation between the measured roughness and the lower or upper limit of the target roughness. For example, when the measured roughness is lower than the lower limit of the target roughness, the roughness difference is the difference between the lower limit of the target roughness and the measured roughness; when the measured roughness is higher than the upper limit of the target roughness, the roughness difference is the difference between the measured roughness and the upper limit of the target roughness; when the measured roughness is within the target roughness range, the roughness difference can be set to zero or set to a small correction value for maintaining the state.

[0068] The historical state correction unit 42 is used to correct the sandblasting compensation amount corresponding to the difference based on the historical sandblasting state of the corresponding surface unit 11. The historical sandblasting state includes at least two of the following: cumulative sandblasting time, cumulative sandblasting energy, previous sandblasting pressure, previous abrasive flow rate, previous nozzle distance, previous jet angle, previous scanning speed, and previous measured roughness.

[0069] In one embodiment, when the cumulative sandblasting energy of a certain surface unit 11 is close to a preset safe energy threshold, and the measured roughness is still lower than the target roughness lower limit, the historical state correction unit 42 limits the increase in sandblasting compensation to avoid damage to the substrate due to continued increase in energy. In another embodiment, when the measured roughness of a certain surface unit 11 increases rapidly compared to the previous measured roughness, even if the current measured roughness has not yet reached the target roughness lower limit, the historical state correction unit 42 can reduce the increment of sandblasting compensation to reduce the risk of roughness overshoot.

[0070] The target blasting energy generation unit 43 is used to generate the target blasting energy for the corresponding surface unit 11 based on the corrected blasting compensation amount. The target blasting energy can be represented by energy level, target energy value, or parameter combination. For example, the target blasting energy can be divided into three levels: low energy, medium energy, and high energy, or it can be represented as an execution parameter vector consisting of blasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed.

[0071] The target sandblasting energy generation unit 43 also receives the predicted roughness output by the roughness trend prediction module 50, and makes forward-looking corrections to the target sandblasting energy based on the prediction deviation. If the predicted roughness will exceed the target upper limit in the next cycle, the target sandblasting energy is appropriately reduced in the current cycle; if the predicted roughness is still below the target lower limit in the next cycle, the target sandblasting energy is appropriately increased in the current cycle. The correction magnitude is related to the prediction confidence interval; the wider the confidence interval, the smaller the correction magnitude, with a conservative strategy being the main approach.

[0072] In a specific example, the target roughness ranges from Ra 6.3 μm to Ra 10.0 μm. When the measured roughness of a certain surface unit 11 is Ra 5.2 μm, the roughness difference calculation unit 41 determines that the surface unit 11 is below the lower limit of the target roughness, and the target sandblasting energy generation unit 43 increases the target sandblasting energy of the surface unit 11. When the measured roughness of another surface unit 11 is Ra 8.0 μm, the target sandblasting energy generation unit 43 reduces the target sandblasting energy of the surface unit 11, causing it to enter a state of sustained sandblasting, low-energy trimming, or skipping the process. When the measured roughness of yet another surface unit 11 is Ra 11.2 μm, the target sandblasting energy generation unit 43 generates a control result to stop sandblasting or reduce the sandblasting energy input. If the roughness trend prediction module 50 predicts that the roughness of a certain surface unit will jump from Ra8.5μm to Ra10.5μm in the next cycle, then even if the current measured value of Ra8.5μm is still within the range, the target sandblasting energy generation unit 43 will reduce the target sandblasting energy of the unit in advance.

[0073] In a preferred embodiment, the system further includes a surface unit memory module 100. The surface unit memory module 100 is used to store the spatial position of each surface unit 11, the sandblasting trajectory 12, the sandblasting execution parameters, the measured roughness, and the historical sandblasting status, and to match the measured roughness obtained by the online roughness detection module 30 with the corresponding surface unit 11.

[0074] The surface unit memory module 100 can be stored using a database, a circular cache, a trajectory queue, or a two-dimensional / three-dimensional mapping table. Each surface unit 11 corresponds to a set of data records, which include the surface unit number, spatial coordinates, target roughness, most recently measured roughness, cumulative blasting time, cumulative blasting energy, most recently blasting pressure, most recently abrasive flow rate, most recently nozzle distance, most recently blasting angle, and most recently scanning speed.

[0075] The surface unit memory module 100 includes a delay registration unit 101. The delay registration unit 101 is used to register the spatial delay or time delay between the sandblasting action and the roughness detection based on the nozzle position, nozzle moving speed, sandblasting action point 13 and roughness detection point 14.

[0076] In one embodiment, the blasting point 13 and the roughness detection point 14 are separated by a distance D along the scanning direction. The nozzle 21 moves along the blasting trajectory 12 at a speed v. The time delay τ between the blasting action and the roughness detection can be determined according to the correspondence between D and v. When the scanning speed v remains stable, the delay time can be determined by τ = D / v. When the scanning speed v changes, the delay registration unit 101 determines the corresponding surface unit 11 based on the real-time speed curve, trajectory integration result, or timestamp queue recorded by the scanning motion unit 26.

[0077] In another embodiment, the delayed registration unit 101 performs spatial matching in the same workpiece coordinate system based on the spatial coordinates of the sandblasting point 13 and the spatial coordinates of the roughness detection point 14. When the roughness detection point 14 falls within the boundary range of a certain surface unit 11, the measured roughness obtained by the online roughness detection module 30 is written into the historical data record of that surface unit 11.

[0078] The delayed registration unit 101 also incorporates a Kalman filter or particle filter subunit for spatiotemporal fusion and smoothing of the registered roughness data. The Kalman filter utilizes the system's state equation and observation equation to optimally estimate the roughness measurement, effectively suppressing sensor noise and registration errors. The particle filter is suitable for nonlinear, non-Gaussian systems, approximating the posterior probability distribution with a set of weighted particles, further improving registration robustness under complex scanning trajectories. The fused roughness data is input as feedback to the blasting energy distribution module 40 and the roughness trend prediction module 50.

[0079] By using the aforementioned delayed registration, the system can avoid mistaking the currently detected roughness for the roughness of the surface unit 11 corresponding to the current position of the nozzle 21, thereby improving the accuracy of closed-loop control.

[0080] In a preferred embodiment, the system further includes an abrasive condition monitoring module 60. The abrasive condition monitoring module 60 includes at least two of the following: an abrasive flow sensor 61, a nozzle inlet pressure sensor 62, a nozzle outlet acoustic sensor 63, and a dust concentration sensor 64.

[0081] The abrasive flow sensor 61 is used to detect the abrasive supply rate or the abrasive throughput per unit time. The nozzle inlet pressure sensor 62 is used to detect the gas pressure at the inlet of nozzle 21. The nozzle outlet acoustic sensor 63 is used to detect jet noise, abrasive impact sound, or abnormal wear sound at the outlet of nozzle 21. The dust concentration sensor 64 is used to detect the dust concentration in the sandblasting area or near the optical path.

[0082] The abrasive condition monitoring module 60 sends the collected abrasive condition information to the adaptive controller 70. The adaptive controller 70 adjusts the target blasting energy or cooperative control commands based on the abrasive condition information.

[0083] For example, when the abrasive flow sensor 61 detects that the actual abrasive flow is lower than the set abrasive flow, the effective impact on the surface is weakened under the same blasting pressure. At this time, the adaptive controller 70 can increase the feeding command of the abrasive flow regulation unit 23, or reduce the scanning speed within a safe range to compensate for the insufficient actual blasting effect.

[0084] When the nozzle inlet pressure sensor 62 detects pressure fluctuations exceeding the set range, the adaptive controller 70 can reduce the target sandblasting energy change rate to avoid overshoot caused by the superposition of pressure fluctuations and roughness feedback adjustment.

[0085] When the nozzle outlet acoustic sensor 63 detects increased nozzle wear characteristics, the adaptive controller 70 can correct the energy mapping relationship of the nozzle 21, reduce the upper limit of the high-energy sandblasting command, or trigger a maintenance prompt.

[0086] When the dust concentration sensor 64 detects that the dust concentration exceeds the set threshold, the adaptive controller 70 can reduce the confidence of the current roughness detection data, use historical data smoothing, delay confirmation, or enter a conservative sandblasting mode to reduce the impact of detection error on sandblasting energy distribution.

[0087] In a preferred embodiment, the system further includes an anomaly protection module 80. The anomaly protection module 80 is communicatively connected to the adaptive controller 70 and is used to control the sandblasting execution module 20 to perform protective actions when the measured roughness exceeds the upper limit of the target roughness, the roughness detection signal is abnormal, the nozzle distance is abnormal, the abrasive flow rate is abnormal, or the nozzle wear is abnormal.

[0088] The protective actions include at least one of reducing sandblasting pressure, reducing abrasive flow rate, increasing nozzle distance, increasing scanning speed, or pausing sandblasting.

[0089] When the measured roughness exceeds the upper limit of the target roughness, the abnormal protection module 80 determines that there is a risk of over-blasting on the corresponding surface unit 11, and controls the adaptive controller 70 to reduce the subsequent sandblasting pressure and abrasive flow rate applied to the surface unit 11, or controls the scanning motion unit 26 to skip the surface unit 11.

[0090] When the output signal of the online roughness detection module 30 is lost, exceeds a reasonable range, or fluctuates continuously beyond a set threshold, the anomaly protection module 80 determines that the roughness detection signal is abnormal and controls the system to enter a low-energy conservative mode. In the low-energy conservative mode, the adaptive controller 70 can reduce the sandblasting pressure, increase the scanning speed, or pause sandblasting and wait for the detection signal to recover.

[0091] When the nozzle distance adjustment unit 24 detects that the distance between the nozzle 21 and the surface of the workpiece 10 is less than the safe distance, the abnormal protection module 80 controls the nozzle distance adjustment unit 24 to increase the nozzle distance, or controls the scanning motion unit 26 to stop moving, so as to avoid the nozzle 21 colliding with the workpiece 10 or the local sandblasting effect being too strong.

[0092] When the abrasive flow sensor 61 detects that the abrasive flow is lower than the set lower limit or higher than the set upper limit, the abnormal protection module 80 controls the abrasive flow regulating unit 23 to adjust the feeding status and reduce the sandblasting pressure or suspend sandblasting according to the degree of abnormality.

[0093] When the nozzle outlet acoustic sensor 63 detects abnormal nozzle wear, the abnormality protection module 80 limits the output of the high sandblasting energy command and prompts the nozzle 21 to be replaced or to be maintained.

[0094] The adaptive control process for the sandblasting process in this embodiment includes the following steps.

[0095] S1. Surface partitioning modeling. The surface partitioning modeling module 90 divides the surface of the workpiece 10 into multiple surface units 11 based on the geometric information of the workpiece 10, the scanning sandblasting trajectory 12, or the detection coordinates, and establishes the target roughness, measured roughness, and historical sandblasting state for each surface unit 11.

[0096] S1a, Pre-optimization of sandblasting strategy. The sandblasting strategy optimization module 110 uses a multi-objective optimization algorithm to solve for the optimal parameter combination based on the overall workpiece size, target roughness range, equipment capacity, and abrasive characteristics, which serves as the global benchmark.

[0097] S2. Sandblasting execution status acquisition. The adaptive controller 70 acquires sandblasting execution status data such as sandblasting pressure, abrasive flow rate, nozzle distance, jet angle, scanning speed, nozzle position, and sandblasting trajectory 12.

[0098] S3. Online roughness detection. The online roughness detection module 30 detects the surface unit 11 after sandblasting, obtains the measured roughness, and sends the measured roughness to the adaptive controller 70.

[0099] S4. Surface unit matching. The surface unit memory module 100 matches the measured roughness with the corresponding surface unit 11 based on the roughness detection point 14, nozzle position, nozzle moving speed, and sandblasting point 13.

[0100] S5. Roughness difference calculation. Roughness difference calculation unit 41 calculates the difference between the target roughness and the measured roughness.

[0101] S5a. Roughness trend prediction. The roughness trend prediction module 50 predicts the roughness at the next moment based on the historical measured roughness sequence of the corresponding surface unit, and generates the prediction deviation and confidence interval.

[0102] S6. Historical Status Correction. The historical status correction unit 42 corrects the sandblasting compensation amount based on the cumulative sandblasting time, cumulative sandblasting energy, previous sandblasting pressure, previous abrasive flow rate, previous nozzle distance, previous spray angle, previous scanning speed, or previous measured roughness.

[0103] S6a, Forward Correction. The target sandblasting energy generation unit 43 makes forward-looking adjustments to the corrected sandblasting compensation amount based on the predicted roughness deviation to obtain the final target sandblasting energy.

[0104] S7. Target blasting energy generation. The target blasting energy generation unit 43 generates the target blasting energy corresponding to the surface unit 11 based on the corrected blasting compensation amount.

[0105] S8. Cooperative control command generation. The adaptive controller 70 generates cooperative control commands for blasting pressure, abrasive flow rate, nozzle distance, injection angle, and scanning speed based on the target blasting energy, predicted roughness correction amount, and the optimal parameter combination provided by the blasting strategy optimization module.

[0106] S8a, Actuator Dynamic Response Identification. The actuator identification unit 121 applies excitation signals to each regulating unit during system startup or periodic maintenance, collects response data, and identifies the transfer function model or state-space model parameters of each regulating unit.

[0107] S8b, Dynamic Response Prediction and Compensation. The dynamic response prediction unit 122 predicts the actual output trajectory of each regulating unit based on the current state and control commands. The look-ahead command pre-compensation unit 123 performs advance correction on the control commands based on the prediction results, causing the slower-responding regulating units to act earlier. The multi-actuator synchronization registration unit 124 uses the slowest-responding regulating unit as a reference and adds a delay queue to the control commands of other regulating units to achieve multi-actuator synchronization. The actuator status monitoring subunit 125 monitors the actual output in real time, triggering step S8a for re-identification when the deviation exceeds a threshold.

[0108] S9. Sandblasting execution. The sandblasting execution module 20 performs sandblasting treatment on the corresponding surface unit 11 according to the dynamically compensated collaborative control command.

[0109] S10, Abnormal Protection Judgment. The abnormal protection module 80 determines whether there is roughness exceeding the limit, roughness detection signal abnormality, nozzle distance abnormality, abrasive flow rate abnormality, or nozzle wear abnormality; if there is an abnormality, it executes at least one of the following protective actions: reducing sandblasting pressure, reducing abrasive flow rate, increasing nozzle distance, increasing scanning speed, or pausing sandblasting; if there is no abnormality, it continues to execute the sandblasting control of the next surface unit 11.

[0110] In one parameter embodiment, the spray angle is 35°, the nozzle distance is 60 mm, and the scanning speed is 20 mm / s. This parameter combination can be used for localized processing scenarios requiring high sandblasting effect per unit area. In this case, the adaptive controller 70 limits the duration of this parameter combination and monitors the cumulative sandblasting energy through the surface unit memory module 100 to reduce the risk of localized over-blasting.

[0111] In another parametric embodiment, the spray angle is 90°, the nozzle distance is 260 mm, and the scanning speed is 600 mm / s. This parameter combination can be used for low-energy trimming of surface units 11 that are already close to the target roughness range. In this case, the system is mainly used to maintain or slightly correct the surface condition to reduce roughness overshoot caused by continued impact.

[0112] In another parameter embodiment, the spray angle is 60°, the nozzle distance is 150 mm, and the scanning speed is 200 mm / s. This parameter combination can be used for medium-energy sandblasting applications. The adaptive controller 70 can use this parameter combination as a reference parameter and coordinately adjust the sandblasting pressure, abrasive flow rate, and scanning speed based on roughness feedback.

[0113] The nozzle distance can be set from 60mm to 260mm. When the nozzle distance is smaller, the velocity attenuation of abrasive particles reaching the workpiece surface is smaller, resulting in stronger sandblasting effect per unit area; when the nozzle distance is larger, jet diffusion increases, reducing the sandblasting effect per unit area. The spray angle can be set from 35° to 90°. When the spray angle is close to 90°, the normal impact component is larger; when the spray angle is smaller, the tangential component increases, suitable for partial surface cleaning or transition area treatment. The scanning speed can be set from 20mm / s to 600mm / s. At lower scanning speeds, the action time per unit area is longer; at higher scanning speeds, the action time per unit area is shorter. The adaptive controller 70 adjusts the above parameters collaboratively based on the target sandblasting energy, rather than adjusting only a single parameter.

[0114] In a parameter embodiment of actuator dynamic response compensation, the identification model of pressure regulating unit 22 is a first-order inertial plus pure delay model, with parameters of gain K = 0.95 MPa / V, pure delay τ = 65 ms, and inertial time constant T = 45 ms. The identification model of scanning motion unit 26 is a second-order oscillation plus pure delay model, with a rise time (10% to 90%) of approximately 100 ms and acceleration / deceleration time of 120 ms. The identification model of abrasive flow regulating unit 23 is a first-order inertial plus pure delay model, with a pure delay τ = 50 ms and inertial time constant T = 80 ms. The multi-actuator synchronization registration unit 124, based on scanning motion unit 26, adds a 55 ms delay queue (120 ms minus 65 ms) to the control commands of pressure regulating unit 22 and a 70 ms delay queue (120 ms minus 50 ms) to the control commands of abrasive flow regulating unit 23. After the above-mentioned delayed registration, when the sandblasting point reaches the target surface unit, the actual output of the three adjustment units simultaneously reaches the target state.

[0115] In another parameter embodiment of rapid energy switching, the target blasting energy of adjacent surface units is switched from a high energy level to a low energy level, requiring the blasting pressure to decrease from 0.6 MPa to 0.3 MPa, the scanning speed to increase from 150 mm / s to 400 mm / s, and the abrasive flow rate to decrease from 1.5 kg / min to 0.8 kg / min. The dynamic response prediction unit 122 predicts the pressure drop curve and the speed rise curve, predicting that the pressure will drop to 0.35 MPa (overshoot of about 0.05 MPa) and the speed will rise to 350 mm / s (still 50 mm / s below the target value) about 100 ms after the switch. At this time, the "low pressure, medium speed" combination may lead to under-blasting. Based on the prediction results, the look-ahead command pre-compensation unit 123 sends a pressure reduction command to the pressure adjustment unit 22 in advance and adds a first-order filter smoothing. At the same time, it sends an acceleration command to the scanning motion unit 26 in advance, so that the moment when the pressure drops to 0.3 MPa and the moment when the speed rises to 400 mm / s are spatially aligned on the same surface unit, avoiding the occurrence of blasting parameter mismatch on that surface unit.

[0116] In the sandblasting treatment of large steel structural components, the workpiece 10 has a large surface area, with significant differences in local corrosion degree and initial roughness. The surface partitioning modeling module 90 divides the surface of workpiece 10 into multiple surface units 11 and sets the initial target sandblasting energy value according to the initial roughness and corrosion state of each surface unit 11. The online roughness detection module 30 detects the measured roughness of each surface unit 11 after the nozzle 21 scans. The sandblasting energy distribution module 40 adjusts the subsequent sandblasting energy based on the measured roughness feedback, so that areas with insufficient roughness receive higher energy input, and areas that have reached the target range receive lower energy input. The roughness trend prediction module 50 monitors the rate of change of roughness of each unit in real time, predicts the value of the next scanning cycle, and reduces the energy in advance if the prediction is about to exceed the limit, effectively avoiding overshoot.

[0117] In the sandblasting process of aerospace parts, the workpiece 10 may have curved surfaces, edges, or thin-walled areas. The surface partitioning modeling module 90 divides the curved surface into units 11 based on the 3D model or point cloud data. The adaptive controller 70 corrects the target sandblasting energy for different curved surface areas based on the nozzle distance and spray angle. For edge or thin-walled areas, the anomaly protection module 80 limits the cumulative sandblasting energy and maximum sandblasting pressure to reduce the risk of localized substrate damage.

[0118] In the surface treatment of composite material molds, the surface of workpiece 10 is highly sensitive to over-blasting. The system can set the target roughness range to a narrower range and increase the detection frequency of the online roughness detection module 30. When the measured roughness approaches the upper limit of the target roughness, the target blasting energy generation unit 43 reduces the blasting energy input in advance to reduce roughness overshoot.

[0119] In large workpiece machining scenarios involving prolonged continuous sandblasting, wear on the pressure regulating valve's seals can lead to a decrease in response speed as running time increases, and increased oil temperature in the hydraulic drive system can cause changes in the acceleration and deceleration characteristics of the scanning motion unit. During execution, the actuator state monitoring subunit 125 continuously monitors the actual response output of each regulating unit and compares it with the predicted output of the dynamic response prediction unit 122. When the actual rise time of the pressure regulating unit 22 is detected to be more than 15% longer than the predicted value, the actuator identification unit 121 is triggered to re-identify the model, updating the pure delay parameter τ and the inertial time constant T. The updated model parameters are adopted by the dynamic response prediction unit 122 and the look-ahead command pre-compensation unit 123, ensuring that the compensation strategy always adapts to the current state of the actuator and guarantees control accuracy during long-term operation.

[0120] Before sandblasting, this invention establishes a control model for each surface unit 11 using a surface partitioning modeling module 90, ensuring that each surface unit 11 has a target roughness, measured roughness, and historical sandblasting state. During sandblasting, the sandblasting execution module 20 executes sandblasting according to the cooperative control commands output by the adaptive controller 70. The online roughness detection module 30 detects the measured roughness of the treated surface after the sandblasting area of ​​the nozzle 21. The surface unit memory module 100 and the delay registration unit 101 accurately match the detection results to the corresponding surface unit 11.

[0121] The sandblasting energy distribution module 40 generates the target sandblasting energy based on the difference between the target roughness and the measured roughness, combined with the cumulative sandblasting time, cumulative sandblasting energy, and the parameters of the previous sandblasting execution. The abrasive condition monitoring module 60 further provides information such as abrasive flow rate, pressure, acoustic condition, and dust concentration to correct the sandblasting energy distribution results. The adaptive controller 70 converts the target sandblasting energy into coordinated control commands for sandblasting pressure, abrasive flow rate, nozzle distance, jet angle, and scanning speed. The sandblasting actuator dynamic response compensation module 120 identifies, predicts, and compensates the dynamic response of each adjustment unit before outputting the results to the sandblasting execution module 20. The anomaly protection module 80 executes protective actions when the roughness exceeds the limit, detection is abnormal, distance is abnormal, flow rate is abnormal, or nozzle wear is abnormal.

[0122] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An adaptive control system for a sandblasting process based on roughness feedback, comprising a sandblasting execution module and an adaptive controller, wherein the sandblasting execution module includes a nozzle, a pressure regulating unit, an abrasive flow regulating unit, a nozzle distance regulating unit, a nozzle angle regulating unit, and a scanning motion unit, characterized in that, It also includes a surface partitioning modeling module, an online roughness detection module, a sandblasting energy distribution module, a roughness trend prediction module, and a sandblasting strategy optimization module; The surface partitioning modeling module is used to divide the surface of the workpiece to be sandblasted into multiple surface units, and to establish the target roughness, measured roughness and historical sandblasting status for each surface unit. The online roughness detection module is located after the sandblasting area of ​​the nozzle, and is used to obtain the measured roughness of the surface unit after sandblasting, and send the measured roughness to the adaptive controller. The sandblasting energy distribution module is used to generate the target sandblasting energy for each surface unit based on the difference between the target roughness and the measured roughness of each surface unit and the historical sandblasting state of the corresponding surface unit. The roughness trend prediction module is used to generate the predicted roughness for the next moment or the next scanning cycle based on the historical measured roughness sequence of each surface unit using a time series prediction algorithm, and send the predicted roughness to the sandblasting energy distribution module for advance correction of the target sandblasting energy. The sandblasting strategy optimization module is used to generate an optimal combination of parameters for sandblasting pressure, abrasive flow rate, nozzle distance, spray angle and scanning speed by using the roughness of each surface unit as a constraint, the shortest total sandblasting time and the minimum total abrasive consumption as optimization objectives, and adopting a multi-objective optimization algorithm. The optimal parameter combination is then sent to the adaptive controller as a reference benchmark for the collaborative control command. It also includes a sandblasting actuator dynamic response compensation module, which is located between the adaptive controller and the sandblasting actuator module. It is used to identify, predict and compensate for the response delay and dynamic response characteristics of each adjustment unit, and generate a coordinated control command after dynamic compensation. Each adjustment unit includes a pressure adjustment unit, an abrasive flow adjustment unit, a nozzle distance adjustment unit, a nozzle angle adjustment unit and a scanning motion unit. The adaptive controller is used to generate coordinated control commands for blasting pressure, abrasive flow rate, nozzle distance, blasting angle, and scanning speed based on the target blasting energy, the predicted roughness, and the preferred parameter combination, and sends the coordinated control commands to the dynamic response compensation module of the blasting actuator; after the dynamic response compensation module of the blasting actuator performs dynamic response compensation on the coordinated control commands, it controls the blasting actuator to perform adaptive blasting processing on the corresponding surface unit; The dynamic response compensation module of the sandblasting actuator includes an actuator identification unit, a dynamic response prediction unit, a look-ahead command pre-compensation unit, and a multi-actuator synchronous registration unit. The actuator identification unit is used to apply excitation signals to each regulating unit during the system startup or maintenance phase, collect the actual response data of each regulating unit, identify the transfer function model or state space model of each regulating unit, and the identification parameters include at least two of the following: delay time, rise time, overshoot, steady-state error, and response bandwidth of each regulating unit. The dynamic response prediction unit is used to predict the actual output trajectory of each control unit in the future multiple control cycles based on the current state of each control unit and the cooperative control command sequence, and send the predicted actual output to the adaptive controller or the look-ahead command pre-compensation unit. The forward-looking command pre-compensation unit is used to perform advance correction or inverse model compensation on the coordinated control command based on the dynamic response prediction results, so that the actual output of each adjustment unit responds in advance to offset the inherent delay. The multi-actuator synchronization registration unit is used to calculate the difference between the response delays of each control unit, and with the control unit with the slowest response as the benchmark, adds a delay queue to the control commands of other control units, so that each control unit can achieve synchronous response at the spatial position of the sandblasting point.

2. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, The excitation signal used by the actuator identification unit includes a step signal, a sinusoidal sweep signal, or a pseudo-random binary sequence signal; the transfer function model is a first-order inertial plus pure delay model or a second-order oscillation plus pure delay model; the state space model is a linear state space model or a Hammerstein model with input nonlinearity.

3. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, The look-ahead command pre-compensation unit employs zero-pole cancellation compensation, lead-lag correction compensation, or model predictive control compensation based on an inverse model. When the target sandblasting energy undergoes a step change, the look-ahead command pre-compensation unit issues a pressure increase or decrease command in advance based on the response delay of the pressure regulation unit, and synchronously adjusts the time sequence of the scanning speed command according to the acceleration and deceleration characteristics of the scanning motion unit.

4. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, The multi-actuator synchronous registration unit uses the scanning motion unit as the reference actuator, calculates the response delay difference between the pressure adjustment unit, abrasive flow adjustment unit, nozzle distance adjustment unit and nozzle angle adjustment unit and the scanning motion unit, and adds the corresponding delay amount to the control command of each adjustment unit, so that the actual output of all adjustment units reaches the target state at the moment when the sandblasting point reaches the target surface unit.

5. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, The dynamic response compensation module of the sandblasting actuator also includes an actuator status monitoring subunit, which is used to monitor the actual response output of each adjustment unit in real time and compare the actual response output with the predicted output of the dynamic response prediction unit. When the deviation between the actual response and the predicted response exceeds a set threshold, the actuator state monitoring subunit triggers the actuator identification unit to re-identify the model in order to update the dynamic model parameters.

6. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, The dynamic response prediction unit is also used to predict the overshoot and oscillation trend of each adjustment unit in the transient process under rapid energy switching conditions; when it is predicted that the transient process overshoot will cause the sandblasting energy to exceed the allowable deviation range of the target sandblasting energy, the forward command pre-compensation unit adds a smoothing constraint to the control command to limit the sandblasting energy change rate to within the safe change rate range.

7. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, The historical sandblasting status includes at least two of the following for the corresponding surface unit: cumulative sandblasting time, cumulative sandblasting energy, previous sandblasting pressure, previous abrasive flow rate, previous nozzle distance, previous spray angle, previous scanning speed, and previous measured roughness.

8. The adaptive control system for sandblasting process based on roughness feedback according to claim 1, characterized in that, It also includes a surface unit memory module, which is used to store the spatial position, sandblasting trajectory, sandblasting execution parameters, measured roughness and historical sandblasting status of each surface unit, and to match the measured roughness obtained by the online roughness detection module with the corresponding surface unit; The surface unit memory module includes a delay registration unit, which is used to register the spatial or temporal delay between sandblasting and roughness detection based on the nozzle position, nozzle moving speed, sandblasting point position, and roughness detection point position.

9. The adaptive control system for sandblasting process based on roughness feedback according to any one of claims 1 to 8, characterized in that, It also includes an anomaly protection module, which is used to control the sandblasting execution module to perform at least one of the following protective actions when the measured roughness exceeds the target roughness upper limit, the roughness detection signal is abnormal, the nozzle distance is abnormal, the abrasive flow rate is abnormal, or the nozzle wear is abnormal: reducing the sandblasting pressure, reducing the abrasive flow rate, increasing the nozzle distance, increasing the scanning speed, or pausing sandblasting.

Images