Controllable distribution control system for shot blasting

Abstract

The invention relates to the technical field of advanced manufacturing, in particular to a controllable distribution control system for shot blasting, which has the advantages that the processing uniformity is improved, so that the shot distribution standard deviation is reduced to 0.03 mm from 0.15 mm of the traditional process, the surface roughness Ra value is reduced by 30%, the process stability is improved, the equipment comprehensive efficiency (OEE) is improved to 92% from 65%, the residual stress fluctuation range is reduced by 50%, the intelligent level is broken through, and the product quality is improved. The process programming of a new workpiece is completed within 20 minutes, the remodeling time is shortened by 80%, the model training time is shortened by 40%, and the energy consumption is reduced: the processing energy consumption of a single piece is reduced by 18% and the carbon emission is reduced by 15% through parameter optimization.

Description

Technical Field

[0001] This invention relates to the field of advanced manufacturing technology, and in particular to a controllable distribution control system for shot peening. Background Technology

[0002] Traditional shot peening processes achieve machining by impacting the workpiece surface with a high-speed stream of shot, but they have the following technical drawbacks:

[0003] 1. Uniformity control challenge: The distribution of projectiles relies on manual experience, and over-spraying (causing surface damage) or under-spraying (affecting the strengthening effect) is prone to occur when machining complex curved surfaces;

[0004] 2. Low parameter coupling: Parameters such as injection pressure, scanning speed, and shot flow rate lack real-time feedback adjustment mechanisms, and environmental changes or equipment wear can lead to a decrease in process stability;

[0005] 3. Insufficient intelligence: The equipment cannot adapt to the rapid changeover requirements of multi-variety, small-batch production modes, and lacks adaptive learning capabilities.

[0006] While existing technologies have proposed shot peening path planning methods based on vision detection, they have not solved the problem of multi-physics coupling control. Therefore, there is an urgent need for a shot peening system that integrates multi-sensor fusion and intelligent control algorithms. Summary of the Invention

[0007] The purpose of this invention is to provide a controllable distribution control system for shot peening, which is suitable for precision machining needs such as surface strengthening, oxide layer removal and stress relief of metal parts in aerospace, automobile manufacturing, mold processing and other fields.

[0008] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0009] A controllable distribution control system for shot peening, comprising:

[0010] a. Multi-degree-of-freedom spray gun actuator, including a robotic arm with at least six rotating joints, and an integrated pressure-adjustable nozzle assembly at the end, with a nozzle flow rate adjustment range of 5-50 kg / min;

[0011] b. 3D scanning module, which uses a combination of structured light sensor and laser displacement sensor, with a scanning frequency ≥500Hz, resolution 0.01mm, and supports direct import of STL format models;

[0012] c. The projectile flow field monitoring unit consists of an array of piezoelectric sensors, covering a 500mm×500mm area in front of the spray gun, with a sampling interval ≤1mm and a data transmission delay <1ms;

[0013] d. Central control unit, equipped with a multi-core DSP processor, running a real-time operating system (RTOS), including:

[0014] The path planning module generates a variable cross-section spiral scanning path based on the workpiece's 3D point cloud data, and supports direct conversion from CAD models.

[0015] A parameter decoupling algorithm library is used to establish a three-dimensional mapping model of injection pressure, projectile velocity, and scanning velocity, with a model update cycle of ≤10ms.

[0016] The feedback compensator uses a sliding mode control algorithm to process sensor data, adjusts the duty cycle of the pulse width modulation (PWM) signal, and controls the response time to <5ms.

[0017] e. Human-machine interface, supports visual editing of processing parameters and historical data backtracking, data storage capacity ≥1TB, supports USB3.0 / Ethernet data export.

[0018] Furthermore, the path planning module performs the following steps:

[0019] (1) Obtain the initial point cloud of the workpiece using a structured light sensor, with a point cloud density ≥ 10 points / mm. 2 ;

[0020] (2) A digital twin model was generated using the Poisson surface reconstruction algorithm, with a model error ≤ 0.05 mm;

[0021] (3) According to the preset coverage requirements (80%-120%), the model is divided into multiple processing areas, and the smoothness of the transition between the area boundaries is ≥C1 continuous.

[0022] (4) Generate a variable pitch helical trajectory in each region, where the pitch Δ satisfies:

[0023]

[0024] Where Vscan is the scanning speed, Tpulse is the pulse period, Nparticle is the number of projectiles per pulse, Dparticle is the projectile diameter, and Kgeom is the geometric correction factor (0.8-1.2).

[0025] (5) Optimize trajectory smoothness by combining the kinematic constraints of the robotic arm, generate joint space motion commands, and ensure that the acceleration mutation rate is ≤5m / s². 3 .

[0026] Furthermore, the piezoelectric sensor array of the projectile flow field monitoring unit adopts a honeycomb layout, and each sensor node includes:

[0027] (1) Piezoelectric ceramic sheet, thickness 0.2mm, resonant frequency 200kHz, sensitivity ≥50mV / N;

[0028] (2) Preamplifier circuit, with adjustable gain range of 40-80dB and noise equivalent pressure ≤1μPa;

[0029] (3) Digital filter bank, including Butterworth bandpass filter (cutoff frequency 50-150kHz) and notch filter (50Hz power frequency interference suppression), stopband attenuation ≥60dB;

[0030] (4) Data fusion module, which realizes spatiotemporal alignment of multi-sensor data through extended Kalman filter algorithm, with spatial resolution error ≤0.5mm and time synchronization accuracy ≤0.1ms.

[0031] Furthermore, the parameter decoupling algorithm library includes:

[0032] (1) The jet pressure-projectile velocity mapping model Vp satisfies:

[0033]

[0034] Where Vp is the projectile velocity, P is the air pressure, ρp is the projectile density, Kp is the nozzle flow coefficient, α is the temperature correction coefficient, and Tamb / Tref is the ambient temperature / reference temperature.

[0035] (2) The scanning speed-coverage relationship model C satisfies:

[0036]

[0037] k is the material absorption coefficient, t is the action time, β is the projectile diameter correction coefficient, and Dpart / Dref is the actual / reference projectile diameter;

[0038] (3) The multi-objective optimization function F satisfies:

[0039] w1·σ residual +w2·ΔE+w3·T cycle +w4·E consum

[0040] Where w1-w4 are weighting coefficients, σresidual is residual stress, ΔE is the change in surface roughness, Tcycle is the single-piece processing time, and Econsum is energy consumption.

[0041] Furthermore, the feedback compensator employs an improved sliding mode control strategy, and its control law is as follows:

[0042]

[0043] Where s(t) is the sliding surface function, Φ is the boundary layer thickness, K is the gain matrix, xd(t) is the desired trajectory, λ is the integral gain, and e(τ) is the tracking error;

[0044] sat() is a saturation function, and its expression is:

[0045]

[0046] Furthermore, it also includes a security protection module, which has:

[0047] (1) Projectile recovery rate monitoring function: when the recovery rate is <95%, an automatic shutdown is triggered, and the shutdown response time is ≤100ms;

[0048] (2) Human proximity detection is achieved by using millimeter-wave radar and infrared sensor fusion sensing. When the safe distance is ≤50cm, an emergency stop is initiated, and the emergency stop time is ≤50ms.

[0049] (3) Dust concentration monitoring, equipped with a laser scattering dust meter, the concentration exceeds the limit (8mg / m³). 3 The dust removal device is activated in conjunction with the system, with a dust removal efficiency of ≥99%.

[0050] Furthermore, the human-computer interaction interface supports:

[0051] (1) Offline programming of machining parameters: the initial process plan is automatically generated by importing CAD models, and STEP / IGES format is supported;

[0052] (2) Online monitoring of projectile flow field distribution heat map, supporting 2D / 3D visualization switching, refresh rate ≥30Hz;

[0053] (3) Historical data backtracking function, storing processing parameters, sensor data, energy consumption records, and data tags including workpiece batch number, operator ID, ambient temperature and humidity, and supporting SQL query.

[0054] Furthermore, it also includes an adaptive learning module, whose workflow is as follows:

[0055] (1) Record the combination of processing parameters and effect evaluation of typical workpieces. The data storage format is HDF5.

[0056] (2) Construct an LSTM neural network model. The input layer includes material properties (Young's modulus, hardness), geometric features (radius of curvature, thickness), and target performance parameters (residual stress, roughness).

[0057] (3) The control strategy is optimized using reinforcement learning algorithms, and the reward function is designed as follows:

[0058] R=α·(1-|σ target -σactual |)+β·(1-R atarget -Ra actual |)-γ·T cycle

[0059] Where α, β, and γ are weighting factors, σtarget / σactual are the target / actual residual stress, and Ratarget / Raactual are the target / actual surface roughness.

[0060] Furthermore, the adaptive learning module adopts a federated learning architecture, supporting model parameter sharing among multiple devices, with the communication period T satisfying:

[0061]

[0062] Where Nnodes is the number of networked devices, Smodel is the model parameter size, Bchannel is the communication channel bandwidth, and η is the compression efficiency (0.6-0.9).

[0063] As an improvement, the beneficial effects of the present invention are as follows:

[0064] 1. Improved processing uniformity: The standard deviation of shot distribution is reduced from 0.15mm in the traditional process to 0.03mm, and the surface roughness Ra value is reduced by 30%;

[0065] 2. Improved process stability: Overall equipment efficiency (OEE) increased from 65% to 92%, and residual stress fluctuation range decreased by 50%;

[0066] 3. Breakthrough in intelligent level: Achieve new workpiece process programming within 20 minutes, reduce changeover time by 80%, and reduce model training time by 40%;

[0067] 4. Reduced energy consumption: Through parameter optimization, energy consumption per piece of processing is reduced by 18%, and carbon emissions are reduced by 15%. Attached Figure Description

[0068] Figure 1 This is a diagram illustrating the overall system architecture of a controllable distribution control system for shot peening according to the present invention.

[0069] Figure 2 This is a layout diagram of the central control unit of the present invention; Detailed Implementation

[0070] To make the content of this invention easier to understand, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings.

[0071] Taking shot peening of automotive drive shafts as an example:

[0072] 1. Workpiece modeling: The 3D point cloud of the drive shaft is acquired by a structured light sensor, and a digital twin model is generated by applying the Poisson reconstruction algorithm. The model error is 0.03mm.

[0073] 2. Path planning: Divide the model into cylindrical segments, flanges, and other regions, and generate variable pitch helical trajectories for each, with a pitch Δ = 0.8 mm (calculated according to the formula);

[0074] 3. Parameter configuration: Set the target residual stress σtarget = 800MPa, surface roughness Ratarget = 1.6μm, and weighting coefficients w1 = 0.4, w2 = 0.3, w3 = 0.2, w4 = 0.1;

[0075] 4. Processing execution: The robotic arm moves along the planned path, the nozzle pressure is dynamically adjusted (4-6 bar), and the scanning speed Vscan = 300 mm / s;

[0076] 5. Real-time monitoring: The piezoelectric array monitors the projectile flow field distribution in real time. When insufficient local coverage is detected, the feedback compensator adjusts the PWM signal duty cycle (from 60% to 75%).

[0077] 6. Quality Assessment: After processing is completed, the system automatically generates a quality inspection report containing stress distribution cloud map and roughness measurement values, and the data is stored in the local database;

[0078] 7. Model Optimization: The processed data was uploaded to the federated learning platform to participate in the global model update, which improved the model accuracy by 15%.

[0079] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A controllable distribution control system for shot peening, characterized in that, include: a. Multi-degree-of-freedom spray gun actuator, including a robotic arm with at least six rotating joints, and an integrated pressure-adjustable nozzle assembly at the end, with a nozzle flow rate adjustment range of 5-50 kg / min; b. 3D scanning module, which uses a combination of structured light sensor and laser displacement sensor, with a scanning frequency ≥500Hz, resolution 0.01mm, and supports direct import of STL format models; c. The projectile flow field monitoring unit consists of an array of piezoelectric sensors, covering a 500mm×500mm area in front of the spray gun, with a sampling interval ≤1mm and a data transmission delay <1ms; d. Central control unit, equipped with a multi-core DSP processor, running a real-time operating system (RTOS), including: The path planning module generates a variable cross-section spiral scanning path based on the workpiece's 3D point cloud data, and supports direct conversion from CAD models. A parameter decoupling algorithm library is used to establish a three-dimensional mapping model of injection pressure, projectile velocity, and scanning velocity, with a model update cycle of ≤10ms. The feedback compensator uses a sliding mode control algorithm to process sensor data, adjusts the duty cycle of the pulse width modulation (PWM) signal, and controls the response time to <5ms. e. Human-machine interface, supports visual editing of processing parameters and historical data backtracking, data storage capacity ≥1TB, supports USB3.0 / Ethernet data export.

2. The controllable distribution control system for shot peening according to claim 1, characterized in that, The path planning module performs the following steps: (1) Obtain the initial point cloud of the workpiece using a structured light sensor, with a point cloud density ≥ 10 points / mm. 2 ; (2) A digital twin model was generated using the Poisson surface reconstruction algorithm, with a model error ≤ 0.05 mm; (3) According to the preset coverage requirements (80%-120%), the model is divided into multiple processing areas, and the smoothness of the transition between the area boundaries is ≥C1 continuous. (4) Generate a variable pitch helical trajectory in each region, where the pitch Δ satisfies: Where Vscan is the scanning speed, Tpulse is the pulse period, Nparticle is the number of projectiles per pulse, Dparticle is the projectile diameter, and Kgeom is the geometric correction factor (0.8-1.2). (5) Optimize trajectory smoothness by combining the kinematic constraints of the robotic arm, generate joint space motion commands, and ensure that the acceleration mutation rate is ≤5m / s². 3 .

3. The controllable distribution control system for shot peening according to claim 1, characterized in that, The piezoelectric sensor array of the projectile flow field monitoring unit adopts a honeycomb layout, and each sensor node includes: (1) Piezoelectric ceramic sheet, thickness 0.2mm, resonant frequency 200kHz, sensitivity ≥50mV / N; (2) Preamplifier circuit, with adjustable gain range of 40-80dB and noise equivalent pressure ≤1μPa; (3) Digital filter bank, including Butterworth bandpass filter (cutoff frequency 50-150kHz) and notch filter (50Hz power frequency interference suppression), stopband attenuation ≥60dB; (4) Data fusion module, which realizes spatiotemporal alignment of multi-sensor data through extended Kalman filter algorithm, with spatial resolution error ≤0.5mm and time synchronization accuracy ≤0.1ms.

4. The controllable distribution control system for shot peening according to claim 1, characterized in that, The parameter decoupling algorithm library includes: (1) The jet pressure-projectile velocity mapping model Vp satisfies: Where Vp is the projectile velocity, P is the air pressure, ρp is the projectile density, Kp is the nozzle flow coefficient, α is the temperature correction coefficient, and Tamb / Tref is the ambient temperature / reference temperature. (2) The scanning speed-coverage relationship model C satisfies: k is the material absorption coefficient, t is the action time, β is the projectile diameter correction coefficient, and Dpart / Dref is the actual / reference projectile diameter; (3) The multi-objective optimization function F satisfies: w1·s residual +w2·ΔE+w3·T cycle +w4·E consum Where w1-w4 are weighting coefficients, σresidual is residual stress, ΔE is the change in surface roughness, Tcycle is the single-piece processing time, and Econsum is energy consumption.

5. The controllable distribution control system for shot peening according to claim 1, characterized in that, The feedback compensator employs an improved sliding mode control strategy, and its control law is as follows: Where s(t) is the sliding surface function, Φ is the boundary layer thickness, K is the gain matrix, xd(t) is the desired trajectory, λ is the integral gain, and e(τ) is the tracking error; sat() is a saturation function, and its expression is:

6. The controllable distribution control system for shot peening according to claim 1, characterized in that, It also includes a security protection module, which has: (1) Projectile recovery rate monitoring function: when the recovery rate is <95%, an automatic shutdown is triggered, and the shutdown response time is ≤100ms; (2) Human proximity detection is achieved by using millimeter-wave radar and infrared sensor fusion sensing. When the safe distance is ≤50cm, an emergency stop is initiated, and the emergency stop time is ≤50ms. (3) Dust concentration monitoring, equipped with a laser scattering dust meter, the concentration exceeds the limit (8mg / m³). 3 The dust removal device is activated in conjunction with the system, with a dust removal efficiency of ≥99%.

7. The controllable distribution control system for shot peening according to claim 1, characterized in that, The human-computer interaction interface supports: (1) Offline programming of machining parameters: the initial process plan is automatically generated by importing CAD models, and STEP / IGES format is supported; (2) Online monitoring of projectile flow field distribution heat map, supporting 2D / 3D visualization switching, refresh rate ≥30Hz; (3) Historical data backtracking function, storing processing parameters, sensor data, energy consumption records, and data tags including workpiece batch number, operator ID, ambient temperature and humidity, and supporting SQL query.

8. A controllable distribution control system for shot peening according to any one of claims 1 to 7, characterized in that, It also includes an adaptive learning module, whose workflow is as follows: (1) Record the combination of processing parameters and effect evaluation of typical workpieces. The data storage format is HDF5. (2) Construct an LSTM neural network model. The input layer includes material properties (Young's modulus, hardness), geometric features (radius of curvature, thickness), and target performance parameters (residual stress, roughness). (3) The control strategy is optimized using reinforcement learning algorithms, and the reward function is designed as follows: R=α·(1-|σ target -σ actual |)+β·(1-|Ra target -Ra actual |)-γ·T cycle Where α, β, and γ are weighting factors, σtarget / σactual are the target / actual residual stress, and Ratarget / Raactual are the target / actual surface roughness.

9. A controllable distribution control system for shot peening according to claim 8, characterized in that, The adaptive learning module adopts a federated learning architecture, supports model parameter sharing among multiple devices, and the communication period T satisfies: Where Nnodes is the number of networked devices, Smodel is the model parameter size, Bchannel is the communication channel bandwidth, and η is the compression efficiency (0.6-0.9).

Images