Intelligent throwing and spraying control method and system for large complex casting

Abstract

The invention discloses an intelligent throwing and spraying control method and system for a large complex casting, and relates to the technical field of casting surface treatment. The method comprises the steps that a shot blasting system and a shot blasting system are initialized, a casting to be cleaned is placed, and preprocessing operation of data collection is carried out; information of a to-be-cleaned casting is collected, a blasting operation path is planned according to the collected information, the planned operation path is optimized by utilizing a genetic algorithm, and a shot blasting system and work scheduling of the shot blasting system are carried out according to the optimized operation path; and errors are calculated according to the shot blasting system and the working condition of the shot blasting system, and the working parameters are subjected to self-adaptive adjustment according to the errors. According to the invention, a three-dimensional model and an online visual inspection technology are combined, automatic partitioning, path planning and parameter adaptive adjustment and optimization are carried out on the surface of the workpiece, and seamless cooperation and real-time closed-loop control of the throwing and spraying module are realized.

Description

Technical Field

[0001] This invention relates to the field of surface treatment technology for castings, and in particular to an intelligent polishing and spraying control method and system for large and complex castings. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In modern industrial production, surface treatment of large and complex castings is a crucial part of the manufacturing process. The surface quality of castings directly affects their subsequent processing performance, service life, and appearance quality. Therefore, efficient and precise cleaning methods are needed to remove impurities such as oxide scale and sand particles from the surface, while achieving the desired surface roughness. Although traditional shot blasting and shot peening systems are widely used in the field of casting surface treatment, they have many limitations and cannot meet the ever-increasing demands for production quality and efficiency.

[0004] Large and complex castings are characterized by intricate geometric shapes, numerous curved surfaces, and many pores. Traditional shot blasting and shot peening equipment based on preset trajectories often suffers from numerous blind spots and low efficiency. Furthermore, parameter adjustments in traditional equipment rely on manual experience and lack real-time feedback and adaptive capabilities. In addition, the coordination between multiple process modules (shot blasting and shot peening) is insufficient, and the switching and coordination effects are inconsistent, resulting in poor cleaning consistency and making it difficult to balance quality and efficiency.

[0005] In summary, existing shot blasting and shot peening systems for large and complex castings have significant shortcomings in collaborative operation, intelligent control, and handling of complex casting surface treatment. There is an urgent need for a solution to overcome these technical challenges, achieve efficient, precise, and intelligent casting surface treatment, meet the stringent requirements of modern industrial production for casting surface quality, improve production efficiency and product quality, and promote technological progress and development in related manufacturing industries. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide an intelligent blasting and spraying control method and system for large and complex castings. By combining three-dimensional models and online visual inspection technology, the workpiece surface is automatically partitioned, path planned, and parameters are adaptively optimized to achieve seamless collaboration and real-time closed-loop control of the blasting and spraying modules.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solution: The first aspect of this invention provides an intelligent blasting and spraying control method for large and complex castings, comprising the following steps: Initialize the shot blasting and shot peening systems, place the castings to be cleaned, and perform preprocessing operations for data acquisition; Information on the castings to be cleaned is collected, and the shot blasting operation path is planned based on the collected information. The planned operation path is optimized using a genetic algorithm, and the operation scheduling of the shot blasting system and shot peening system is performed based on the optimized operation path. The error is calculated based on the working conditions of the shot blasting system and the shot peening system, and the working parameters are adaptively adjusted according to the error.

[0008] Furthermore, the specific steps for initializing the shot blasting system and shot peening system are as follows: Configure and install the hardware for initializing the shot blasting and shot peening systems; Configure and initialize the software for the initial shot blasting system and shot peening system; Check the operating status of the shot blasting system and shot peening system.

[0009] Furthermore, the specific steps for the preprocessing operation of placing the casting to be cleaned and collecting data are as follows: Model the casting to be cleaned and mesh it; Priority weights are calculated and assigned according to the characteristics of the areas to be operated on the casting; The shot blasting system and shot peening system are performed online for attitude recognition and registration based on the placed castings; The placed castings are divided into work sub-areas.

[0010] Furthermore, areas with high curvature, deep blind spots, and narrow gaps are given high weight, while flat and easily accessible areas are given low weight.

[0011] Furthermore, the specific steps for dividing the placed castings into working sub-areas are as follows: Divide into multiple sub-regions based on priority weights; The cleanup method is mapped according to the characteristics of each sub-region, and the switching point and cleanup method are recorded.

[0012] Furthermore, the specific steps for planning the spraying operation path based on the collected information are as follows: Construct a sub-region node diagram based on the divided sub-regions, with each node being the center of the grid. Feasible paths are generated based on the combined distance of edge weights and the incident angle penalty.

[0013] Furthermore, the specific steps for calculating the error based on the working conditions of the shot blasting system and the shot peening system, and for adaptively adjusting the working parameters based on the error, are as follows: Collect images of the casting surface after cleaning and inspect the cleaning effect; Error calculation is performed based on the test results; Based on fuzzy logic or reinforcement learning methods, the current working parameters are adjusted according to the error.

[0014] A second aspect of the present invention provides an intelligent blasting and spraying control system for large and complex castings, comprising: The workpiece 3D modeling and recognition module is configured to initialize the shot blasting system and shot peening system, place the casting to be cleaned, and perform preprocessing operations for data acquisition. The work scheduling module is configured to collect information on the castings to be cleaned, plan the shot blasting operation path based on the collected information, optimize the planned operation path using a genetic algorithm, and schedule the work of the shot blasting system and the shot peening system based on the optimized operation path. The feedback adjustment module is configured to calculate the error based on the working conditions of the shot blasting system and the shot peening system, and to adaptively adjust the working parameters according to the error.

[0015] A third aspect of the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed the steps of the intelligent blasting control method for large and complex castings as described in the first aspect of the present invention.

[0016] A fourth aspect of the present invention provides a computer device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the intelligent blasting control method for large and complex castings as described in the first aspect of the present invention.

[0017] The above one or more technical solutions have the following beneficial effects: This invention discloses an intelligent shot peening and shot blasting control method and system for large and complex castings. It achieves efficient surface treatment of complex castings by dividing and planning the casting operation area. Under the intelligent scheduling and control strategy design of the shot peening and shot blasting systems, collaborative operation between shot blasting and shot blasting is realized. Through real-time feedback and adaptive parameter adjustment based on reinforcement learning, the efficiency and quality of casting surface treatment are improved.

[0018] This invention, by introducing an intelligent blasting control algorithm, achieves efficient and precise cleaning of the surfaces of large and complex castings, significantly improving surface treatment quality and operational consistency. Combining workpiece geometry with cleaning requirements, this invention intelligently generates blasting paths and dynamically optimizes operational parameters, reducing manual intervention and improving the system's adaptability and robustness. Through multi-module collaborative scheduling and cleaning load balancing, this invention effectively alleviates operational bottlenecks, improving equipment utilization and overall operational efficiency. Supporting online identification, quality feedback, and intelligent adjustment, this invention possesses good scalability and industrial application value, and is suitable for surface treatment operations on castings with various complex morphologies.

[0019] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of the intelligent polishing and spraying control method for large and complex castings in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the generation of the partitioning and path diagram structure of the complex casting three-dimensional model in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the generation of the partitioning and path diagram structure of the complex casting three-dimensional model in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of parameter adaptive optimization based on cleaning quality feedback in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of parameter adaptive optimization based on cleaning quality feedback in Embodiment 1 of the present invention; Figure 6 This is a schematic diagram of the intelligent blasting and spraying control system for large and complex castings in Embodiment 2 of the present invention. Detailed Implementation

[0022] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0024] Example 1: Embodiment 1 of the present invention provides an intelligent polishing and spraying control method for large and complex castings, such as... Figure 1 As shown, the system first initializes the hardware, software, and corresponding parameters, performs 3D scanning and modeling of the workpiece, and collects data. Then, surface areas are divided based on geometric features, and the cleaning path is optimized using intelligent algorithms. Next, a collaborative control module decides the order of blasting and spraying, prioritizing blasting and spraying operations through cleaning method mapping, and dynamically adjusting parameters. During the operation, the system's quality assessment module continuously receives quality feedback and evaluates the results. Based on the principle of determining whether quality standards are met, a reinforcement learning model is used to adaptively adjust the cleaning parameters, thus forming a complete closed-loop optimization control mechanism.

[0025] Specifically, the following steps are included: S1: Initialize the shot blasting and shot peening systems, place the castings to be cleaned, and perform preprocessing operations for data acquisition.

[0026] S101: Initialize the shot blasting system and shot peening system.

[0027] S1011: Configure and install the hardware for initializing the shot blasting system and shot peening system.

[0028] In one specific implementation, before preparing and initializing the intelligent blasting control system, it is necessary to complete the configuration and installation of hardware devices to ensure that the system has complete functional modules and a stable operating environment.

[0029] Shot blasting system hardware configuration and installation: Select a shot blasting machine suitable for processing large and complex castings, ensuring that parameters such as the shot blasting head's speed and shot flow rate are adjustable. When installing the shot blasting machine, ensure its working area is flat and stable to avoid vibrations during operation due to uneven ground. Adjust the shot blasting head angle and blasting range to cover all surface areas of the casting. Conduct a no-load test of the shot blasting machine to check if the equipment is operating normally, if the shot conveying system is smooth, and if the shot blasting head rotates smoothly.

[0030] Shot peening system hardware configuration and installation: Select a high-pressure shot peening system, ensuring that parameters such as shot peening pressure and shot flow rate are adjustable. When installing the shot peening machine, ensure that the position and angle of its shot peening head can be flexibly adjusted according to the shape of the casting. Adjust the shot peening pressure and range of the shot peening machine to ensure that the shot peening can evenly cover the casting surface, avoiding insufficient or excessive shot peening in certain areas. Conduct a no-load operation test of the shot peening machine to check whether the equipment is operating normally and whether the shot peening system can stably output the set pressure and flow rate.

[0031] S1012: Configure and initialize the software for the initialization of the shot blasting system and shot peening system.

[0032] In one specific implementation, after the hardware equipment is installed, the software system needs to be configured and initialized to ensure that the system can operate normally and realize the intelligent throwing and spraying control function.

[0033] Configure the data acquisition program, set the sensor's sampling frequency and data format, and ensure that sensor data can be read in real time.

[0034] The software system in this embodiment can filter the collected data to remove noise interference and improve data accuracy. It also stores the collected sensor data in a database for subsequent data analysis and processing.

[0035] A large amount of surface treatment data for castings was collected, including information such as the shape, size, surface roughness, and material of the castings, as well as the corresponding shot blasting and peening parameters and treatment effects. Deep learning algorithms (such as convolutional neural networks) were used to train the collected data to construct a mathematical model for casting surface treatment. Through cross-validation and testing, the model parameters were optimized to improve its accuracy and generalization ability.

[0036] S1013: Check the operating status of the shot blasting system and shot peening system.

[0037] In one specific implementation, the equipment operating status is checked to ensure normal communication. Initial operating parameters for the shot blasting and shot peening machines are set, including shot blasting speed, shot peening pressure, and shot flow rate. Parameters for the collaborative control module are initialized, including the shot blasting and shot peening sequence and parameter adjustment step size. Parameters for the quality assessment module are initialized, including the threshold values ​​for the image processing algorithm and quality assessment standards.

[0038] Set initial operating parameters and initialize the collaborative control and quality assessment modules. Perform functional tests to verify the functionality of the data acquisition, collaborative control, and quality assessment modules. Conduct performance tests to evaluate processing efficiency and surface quality, ensuring stable system operation.

[0039] S102: Preprocessing operation for placing the casting to be cleaned and collecting data.

[0040] S1021: Model the casting to be cleaned and mesh it.

[0041] In one specific implementation, a CAD model or point cloud is processed on the casting to be cleaned, and a triangular mesh reconstruction (such as Poisson reconstruction) is performed on the CAD model or point cloud. The curvature index of each triangular facet is calculated using the rate of change of normal.

[0042] In 3D computer graphics, triangular facets are a basic geometric structure used to represent the surface of an object. They are composed of multiple triangles pieced together to form a discretized approximation of a complex curved surface. In this embodiment, they are used to perform mesh modeling of the casting surface for subsequent curvature calculation, region division, and path planning.

[0043] S1022: Calculate and assign priority weights according to the characteristics of the areas to be operated on the casting.

[0044] In one specific implementation, areas with high curvature, deep dead angles, and narrow gaps are assigned high weights, while flat and easily accessible areas are assigned low weights. A weighted map is generated based on the importance and accessibility of the regions.

[0045] Among them, areas of high curvature are regions on the casting surface with drastic local geometric changes (such as grooves, edges, and inner walls of holes). Specifically, based on the rate of change of the normal vector of each triangular facet in the network model, when the rate of change is greater than or equal to 0.5 mm... -1 For high curvature, less than 0.1mm -1 It has a low curvature.

[0046] The deep dead zone is a hidden area that is difficult to directly cover with the shot peening jet (such as deep holes, internal cavities, and the bottom of multi-level steps). Specifically, the realization analysis method is used to emit multiple rays from the center of the area, detect the number of collisions with the casting surface, and calculate the dead zone index D = 1 - number of unobstructed rays / total number of rays. When D is greater than or equal to 0.7, it is a high dead zone; when it is less than 0.3, it is an easily accessible area.

[0047] Narrow gaps are elongated spaces with a width smaller than the effective range of the shot peening medium (such as gaps between cooling fins or sealing grooves). Specifically, a minimum channel width detection method is used to calculate the minimum Euclidean distance within the area, i.e., the shortest distance between adjacent facets. The diameter of the pellets is a narrow area.

[0048] The flat, easily accessible area refers to a surface with low curvature, no obstructions, and which can be covered in a straight line by the shot peening jet (such as a flat plate or an outer cylindrical surface). Specifically, the curvature is <0.1mm. -1 D<0.3; The diameter of the pellets should be within a flat and easily accessible area.

[0049] S1023: Perform online attitude recognition and registration of the shot blasting system and shot peening system based on the placed casting.

[0050] In one specific implementation, the casting to be cleaned is subjected to online real-time point cloud acquisition, and the blasting robotic arm is preset to scan the trajectory or a fixed multi-view camera is used to acquire the workpiece point cloud.

[0051] Then, the ICP (Iterative Closest Point) algorithm was used to register the real-time point cloud with the offline mesh model, and the attitude transformation matrix was solved. .

[0052] Update the actual position of each grid cell in the world coordinate system.

[0053] S1024: Divide the placed castings into working sub-areas.

[0054] S10241: Divided into multiple sub-regions according to priority weight, specifically N sub-regions.

[0055] Specifically, starting with the triangle with the highest weight, sub-regions are generated by aggregating them according to adjacency and weight difference threshold (Δw≤2). Isolated high-weight regions are then separately divided into sub-regions. Morphological operations are used to smooth the boundaries of the sub-regions to generate the final partitioning result.

[0056] S10242: Map the cleanup method according to the characteristics of each sub-region, and record the switching point and cleanup method.

[0057] In one specific implementation, the cleaning method is mapped according to the characteristics of each sub-region. Specifically, the mapping relationship is as follows: In high curvature regions, shot peening is prioritized because shot peening is prone to failure due to collision and obstruction. High-pressure jets can accurately penetrate complex structures, avoiding energy loss caused by shot rebound. In flat regions, shot peening is prioritized because shot peening can quickly cover low curvature surfaces by utilizing its advantages, with a wide-area, high-speed blasting capability that is 3-5 times more efficient than shot peening, and it does not require a high-pressure air source, resulting in lower costs. The module switching points and corresponding sub-region lists are then recorded.

[0058] S2: Collect information about the casting to be cleaned, plan the shot blasting operation path based on the collected information, optimize the planned operation path using a genetic algorithm, and schedule the shot blasting system and shot peening system based on the optimized operation path.

[0059] S201: Plan the path for the spraying operation based on the collected information.

[0060] S2011: Construct a sub-region node diagram based on the divided sub-regions, with each node being the center of the grid.

[0061] S2012: Generate feasible paths based on edge weight combined distance and incident angle penalty.

[0062] Initial use The path generation algorithm generates feasible paths, specifically: First, calculate the edge weighted total distance, which is the weight of the edge connecting two sub-region nodes. The calculation formula is as follows: , .

[0063] The Euclidean distance between the centers of two sub-regions is denoted as the physical distance. Clean up the difference in priority and weight. Weight of target sub-region With current sub-region rights The difference, and These are the weighting coefficients.

[0064] Then, the incident angle penalty and corresponding penalty function are set. When the angle θ between the jet stream and the surface normal vector is too large, the cleaning effect decreases (e.g., (The rebound rate of the material surges), such edges need to be penalized in path planning.

[0065] Penalty function: .

[0066] in, For the penalty function, This is a hyperparameter.

[0067] Use afterwards The path generation algorithm generates feasible paths. It plans a spraying path from the starting point to the ending point, making the path short (reducing the robot arm's movement time); it prioritizes cleaning high-weight areas (such as deep holes and narrow slits).

[0068] Step 1: Divide the surface of the casting into sub-regions. Divide the surface of the casting into multiple small blocks (sub-regions), each with a center point. Each sub-region has a weight value (high weight = requires more intensive cleaning, such as areas with high curvature).

[0069] Step 2: Connect sub-regions. If two sub-regions are adjacent (e.g., A and B can be moved directly between them), connect them with a line (edge). Each line has two attributes: distance (the physical length from A to B) and incident angle (the angle between the jet stream and the surface normal).

[0070] Step 3: The overall score for each path = distance score + priority score + angle penalty score.

[0071] Step 4: Initial path generation, using... The algorithm starts from the starting point, prioritizes paths with lower total scores, and automatically skips invalid paths with an angle greater than 60°.

[0072] Step 5: Optimize the path. Use a genetic algorithm to randomly generate multiple paths, retain the paths with low total scores and high coverage, swap the order of sub-regions or perform local replanning, and select the route with the lowest total score that covers all high-weight regions.

[0073] like Figure 2 The diagram illustrates the region segmentation and path graph generation of a complex 3D casting model. Based on geometric information such as curvature distribution and occlusion locations, the system automatically divides the surface into several sub-regions and maps them to a graph structure. Nodes represent areas to be cleaned, and edges represent the relationships between work paths. Intelligent algorithms (such as genetic algorithms and heuristic search) perform path optimization on this graph structure to generate the blasting trajectory.

[0074] S202: Optimize the planned job path using a genetic algorithm.

[0075] Genetic algorithm (GA) optimization is performed on the path: crossover and mutation generate multiple candidate paths, and the best one is retained.

[0076] S2021: Initial population generation. First, use... The algorithm generates 5 initial paths that satisfy the basic constraints (covering all key areas and having acceptable incident angles).

[0077] Then, 45 feasible paths are randomly generated to ensure population diversity; All paths must pass validity checks: connectivity checks and critical area coverage verification.

[0078] S2022: Fitness Assessment Calculate three metrics for each path: (1) Total path cost = distance traveled × 0.7 + priority weight difference × 0.3 + incident angle penalty.

[0079] (2) Key area coverage rate = Number of covered key areas / Total number of key areas.

[0080] (3) Process quality score = incident angle pass rate × 0.6 + parameter stability × 0.4, final fitness value = 1 / (1+total cost) + 0.5×coverage + 0.3×quality score.

[0081] S2023: Select a path.

[0082] Retain the 5 elite paths with the highest fitness; For the remaining paths, a tournament selection process is used (5 paths are randomly selected each time, and the top 2 are chosen). A total of 30 paths were selected to enter the next generation.

[0083] S2024: Perform an interleaving operation on the selected path.

[0084] Pair the selected paths together.

[0085] Using the sequential crossover (OX) method: a) Randomly select two intersection points b) Retain the segment between the two intersections of parent A. c) Fill the remaining nodes of parent B with the missing nodes in the original order; The crossover probability is set to 85%.

[0086] S2025: Perform a mutation operation on the selected path.

[0087] Implement three mutation methods: (1) Exchange Mutation (30% probability): Randomly swap the positions of two sub-regions. (2) Insertion mutation (40% probability): Randomly insert a missed key subregion. (3) Local optimization mutation (30% probability): Perform the following on the 5 edges with the highest cost. Re-planning; The overall mutation probability was controlled at 15%.

[0088] S2026: Set the iteration termination condition.

[0089] Stop when any of the following conditions are met: The optimal solution has been improved by less than 1% over 15 consecutive generations. Population similarity >75%; The maximum number of iterations (default 100 generations) has been reached. S2027: Output the optimal path.

[0090] For the final optimal path, perform the following: Smooth B-spline trajectories; Process parameter mapping (each sub-region matches the corresponding shot blasting / peening parameters); Collision detection verification; Generate executable G-code.

[0091] S203: Schedule the work of the shot blasting system and shot peening system according to the optimized work path.

[0092] In one specific implementation, the shot blasting system and the shot peening system are each considered as an agent, and each sub-region competes for a time window according to the priority of the shot blasting agent and the shot peening agent. The competition for the time window between the shot blasting system and the shot peening system is a mechanism that uses a dynamic scheduling algorithm to compete for the processing time slots of each sub-region.

[0093] The shot blasting operation uses the following formula: ( (1)。

[0094] Where X represents the current candidate solution. These are randomly selected candidate solutions. This is the current globally optimal solution.

[0095] 1. Methods for generating and obtaining candidate solutions: (1) Generation of the current candidate scheme X: Parent individuals are selected from the population through a selection operation. The roulette wheel selection method is used, and the probability of each individual being selected is proportional to its fitness.

[0096] (2) Random candidate scheme Obtaining X: Randomly select an individual different from X from the current population. To ensure diversity, random sampling without replacement is used. Set a minimum Hamming distance constraint to ensure... It is sufficiently different from X.

[0097] (3) Global Optimal Solution Maintenance: Record the individual with the highest fitness during initialization. After each generation iteration, compare the new individual with the current best. Employ an elite preservation strategy to ensure the best individual is not corrupted. Establish an optimal solution archive to save historical optimal solutions.

[0098] 2. Decision-making mechanism for the plan: (1) Update the formula analysis: .

[0099] Where λ is the random perturbation coefficient (default 0.5) and μ is the optimal guidance coefficient (default 0.3). The path node sequence is then weighted and recombined.

[0100] (2) Specific decision-making steps: a) Calculate the difference components using the symmetric difference of the path node sequence and the key sub-region coverage difference: Random difference components = , Optimal difference component = .

[0101] b) Weighted fusion: Retain 80% of the original nodes in X. From Randomly introduce 15% of new nodes. Force the introduction of 5% of critical nodes.

[0102] c) Validity verification: Check path connectivity, verify key sub-region coverage, ensure no duplicate nodes, and check incident angle constraints.

[0103] (3) Adaptive adjustment strategy: early stage (first 20 generations): increase λ (0.7) to promote exploration; middle stage (20-60 generations): balance λ (0.5) and μ (0.3); late stage (after 60 generations): increase μ (0.5) to enhance convergence.

[0104] The shot peening operation uses the following formula: (2).

[0105] in, For the amplitude of local disturbance, It is a standard normally distributed random vector.

[0106] The central scheduler allocates runtime slots and triggers fast-switch mechanisms based on the current queue and real-time load.

[0107] 1. Runtime segment allocation principles: (1) Dynamic priority calculation: Basic priority: Based on the pre-set cleaning method mapping rules (shot peening priority for high curvature areas, shot blasting priority for flat areas).

[0108] Real-time adjustment factor: a) Device ready status (idle device + 0.2 priority weight).

[0109] b) Sub-region urgency (with a weight of +0.3 if the remaining processing time is less than the threshold).

[0110] c) Load balancing factor (weighted by 0.1 when the current load rate is >80%).

[0111] (2) Time window allocation algorithm: First, the timeline is divided into time slots with a 10ms granularity. The following decision-making process is executed when allocating each slot: a) Check the preset process type (blasting / shot peening) of the sub-area to be processed.

[0112] b) Calculate the real-time priority score for each available device: Score = Base Priority + Σ Adjustment Factor.

[0113] c) Select the device-sub-area combination with the highest score and assign it to the current slot.

[0114] The conditions for triggering the quick-change mechanism are then determined: a) When the process type of adjacent tanks changes.

[0115] b) Quick-change delay is controlled to ≤50ms.

[0116] 2. Execution rules for shot blasting and shot peening operations: (1) Basic execution order: Strictly adhere to the cleanup method mapping table: a) Flat area (curvature < 0.1 mm) -1 ): Shot blasting → Quality inspection → (If it fails, additional shot blasting is required).

[0117] b) Complex areas (curvature ≥ 0.5 mm) -1 ): Shot peening → Quality inspection → (If unqualified, adjust parameters and re-peel).

[0118] c) Transition zone (0.1≤curvature<0.5): Parallel shot peening / blasting requests are initiated, and the right to execute is competed based on real-time scores.

[0119] (2) Multiple processing mechanism: Single processing benchmark: a) Shot blasting: Fixed duration 300ms (adjustable).

[0120] b) Shot peening: Calculated dynamically based on area (50 ms / cm) 2 ).

[0121] Iteration processing conditions: a) Quality inspection failed to meet standards (oxide scale residue >5% or Ra deviation >0.5μm).

[0122] b) Remaining number of processing attempts < maximum set value (default 3 times).

[0123] c) Equipment load rate <90%.

[0124] 3. Patent features of scheduling logic: (1) Priority Decision Tree: Is it a critical path node? (Yes → +0.2), Curvature ≥ 0.5? (Yes → Shot peening +0.3), Curvature < 0.1? (Yes → Shot blasting +0.3), Equipment idle? (Yes → +0.1) (2) Exception handling mechanism: Conflict resolution: When shot blasting / peening requests are tied, the following method shall be used for adjudication: a) Sub-region ID in ascending order.

[0125] b) Device ID in descending order.

[0126] c) Random numbers.

[0127] Timeout handling: If a single operation times out by 150%, a device switch will be triggered immediately.

[0128] like Figure 3As shown, the scheduling module collects the cleaning status and module load of each work area in real time, and dynamically adjusts the module switching strategy according to the adaptability of shot blasting and shot peening and the regional priority, so as to realize the coordinated operation and cycle synchronization of shot blasting and shot peening modules, significantly improving the overall operation efficiency and equipment utilization. Specifically, the entire scheduling system starts to operate from the input module, receiving sensor data from each work area in real time, including the three-dimensional coordinates of the sub-area collected by a high-precision laser scanner (accuracy up to ±0.1mm), the operating status parameters of the shot blasting machine and shot peening equipment (such as current and pressure values), and the real-time pose information of the robotic arm. These raw data first enter the data processing center for preprocessing, eliminating sensor noise interference through the Kalman filter algorithm, and transforming the local coordinate system of all equipment to the global coordinate system to ensure the consistency of spatial positioning. The system then extracts key features, including the sub-area curvature value calculated based on the rate of change of the triangular mesh normal vector, and the dynamic priority weight map generated by combining curvature, dead angle index and gap width. The core of the collaborative scheduling logic adopts a multi-condition decision mechanism, when the sub-area curvature is detected to be ≥0.5mm. -1 The system automatically prioritizes shot peening requests, while defaulting to shot blasting for flat areas with low curvature. In special cases where equipment load is below 85%, the system initiates concurrent shot blasting and shot peening requests, using a scoring mechanism to determine the optimal solution. This scoring system comprehensively considers process matching (60% weight), sub-area priority (30% weight), and equipment readiness (10% weight), dynamically scheduling in 10ms increments. When different process requirements conflict, the system makes rapid decisions according to preset arbitration rules, ensuring process switching latency is strictly controlled within 50ms. This process is precisely synchronized via a high-speed EtherCAT bus. The quality assessment module acquires processed surface images using an industrial camera (50μm / pixel resolution) and measures surface roughness using a laser profilometer. The system uses the Otsu threshold segmentation algorithm to calculate oxide scale residue, calculates the arithmetic mean roughness Ra value using Gaussian filtered profile data, and simultaneously calculates the actual cleaning coverage of key sub-areas. To meet the standards, three core indicators must be met simultaneously: residual oxide scale rate ≤5%, roughness Ra value within ±0.5μm of the target value, and critical sub-area coverage ≥98%. For areas that do not meet the standards, the system will automatically mark them and trigger parameter adjustment strategies, including increasing shot peening pressure by 0.1MPa or shot blasting speed by 5%, while reallocating the processing time window for rework.

[0129] The output module of the entire system generates a PDF report containing cleaning efficiency curves, residual oxide scale distribution heat maps, and equipment utilization statistics. All process data is stored in an SQL database, including detailed process parameters, quality inspection results, and equipment operation logs.

[0130] S3: Calculate the error based on the working conditions of the shot blasting system and the shot peening system, and adaptively adjust the working parameters according to the error.

[0131] like Figure 4 As shown, images of the cleaning area are acquired through industrial vision sensors, and indicators such as surface roughness and residue rate are extracted. Reinforcement learning algorithms (such as DQN) are used to intelligently adjust core parameters such as shot blasting speed and shot peening pressure, thereby continuously improving surface consistency and cleaning accuracy.

[0132] S301: Collect images of the casting surface after cleaning and perform cleaning effect detection.

[0133] Specifically, the process involves collecting surface images and roughness data of the casting after cleaning; calculating the residual oxide scale coverage, roughness deviation, and cleaning uniformity to obtain the comprehensive error; dynamically adjusting the shot blasting speed and shot peening pressure based on fuzzy logic or deep reinforcement learning algorithms; and iteratively optimizing until the error is below a set threshold.

[0134] S3011: Send the end-effector trajectory and corresponding module commands (shot blasting / peening start, parameters) to the robot controller.

[0135] S3012: The robotic arm moves along the trajectory and performs cleaning.

[0136] In this embodiment, the robotic arm and the shot peening system work closely together through a multi-layered collaborative control architecture. The robotic arm, acting as the execution terminal, has its end effector connected to both the shot peening head and the shot blasting head via a quick-change interface. This quick-change mechanism uses a hybrid pneumatic-electric connection, enabling tool switching within 80 milliseconds. The control system employs a distributed architecture, where the main controller establishes communication connections with both the robotic arm driver and the shot peening equipment controller simultaneously via real-time Ethernet (EtherCAT protocol), achieving μs-level time synchronization.

[0137] The motion control commands for the robotic arm are generated uniformly by a central scheduler, including precise timing coordination of pose trajectory and process trigger signals. Specifically, when the robotic arm moves to the target sub-area, the scheduler simultaneously issues a process start command, which includes parameters such as shot blasting speed (adjustable from 12-30 m / s) or shot peening pressure (adjustable from 0.4-0.8 MPa). During operation, the robotic arm's six-dimensional force sensor monitors contact force feedback in real time. When an abnormal collision is detected (force value exceeding the 50N threshold), the shot blasting power output is immediately cut off through a safety circuit.

[0138] The shot peening system uploads its operating status (such as shot flow rate and compressed air pressure) to the scheduler in real time via a dedicated feedback channel. The system employs a predictive-correction control strategy, pre-calculating the coverage area of ​​the shot peening stream during the robotic arm trajectory planning stage. When the actual processing effect is found to be substandard, it automatically generates a composite command including pose fine-tuning and parameter correction. This tightly coupled control method enables tool center point (TCP) positioning accuracy to reach ±0.05mm in cleaning operations on complex curved surfaces such as turbine blades, and the response time for process parameter adjustments does not exceed 20ms, achieving precise coordination between spatial motion and surface treatment processes.

[0139] S3013: Use a binocular camera or laser profiler to acquire images of the cleaned surface.

[0140] S3014: Extract residual oxide scale and roughness indicators using Otsu threshold segmentation and bandpass filtering algorithm.

[0141] 1. Extraction of residual oxide scale: The system first acquires images of the casting surface using an industrial camera (50 μm / pixel resolution). After converting the RGB images to grayscale, the optimal grayscale threshold is automatically determined using the Otsu thresholding algorithm. This algorithm divides pixels into oxide scale regions (foreground) and matrix regions (background) by maximizing inter-class variance. For cases with uneven illumination, adaptive histogram equalization preprocessing is performed before applying local Otsu segmentation (15×15 pixels window size). After segmentation, morphological closing operations (3×3 circular structuring elements) are used to eliminate small holes. Finally, the oxide scale coverage is calculated as a quantification indicator (number of oxide scale pixels / total number of pixels × 100%).

[0142] 2. Roughness measurement: Surface profile curves were acquired using a laser profilometer (sampling interval 10μm). Long-wave and short-wave interference components were first eliminated using a 0.8-25μm bandpass filter. The filtered profile data were then used to calculate the following parameters according to ISO 4287 standard: Arithmetic mean roughness Ra: the arithmetic mean of the absolute values ​​of the filtered profiles; Maximum peak-valley height Rz: the average difference between the five highest peaks and five lowest valleys within the sampling length. Three parallel profile lines (1mm apart) were taken along the direction perpendicular to the machining texture during measurement, and the average value was used as the final result.

[0143] 3. Pass / Fail Judgment: The system is configured with dual thresholds: (1) Oxide scale residue rate ≤3% (automotive parts) or ≤5% (ordinary castings).

[0144] (2) The roughness Ra value is within the range of ±0.8 μm of the target value (e.g., if the target Ra = 3.2 μm, then the acceptable range is 2.4-4.0 μm).

[0145] Rework is triggered when the following occurs: a continuous area of ​​oxide scale exceeding the standard > 4mm. 2 Ra value exceeds the limit and is accompanied by Rz value exceeding 4 times Ra; two tests of the same sub-region fail.

[0146] S302: Calculate the error based on the test results.

[0147] The errors include the error between the target cleaning rate and the actual cleaning rate, and the deviation between the target roughness and the actual roughness.

[0148] S3015: Return the detection results to the adaptive algorithm module.

[0149] S303: Adjust the current working parameters based on the error using fuzzy logic or reinforcement learning methods.

[0150] In one specific implementation, if fuzzy logic is used, the process involves inputting the error, applying Mamdani fuzzy rules, and outputting the incremental projectile velocity and shot peening pressure. If DQN reinforcement learning is used, the process involves using the error and historical actions as states, and outputting discrete parameters to adjust the actions.

[0151] The adjusted working parameters are sent out and written back to the PLC controller and the shot peening pump / shot blasting machine control system. If the adjustment exceeds the safety threshold, an alarm will be automatically triggered and the operation will be suspended.

[0152] S304: For each sub-region, if the current monitoring error is within the allowable range, mark it as "Completed". Repeat the adaptive working parameter adjustment steps until all N sub-regions have been processed.

[0153] S305: Summarize the data and generate a quality report.

[0154] S4: Clean up the collection, storage, and report generation of work data, as well as subsequent maintenance and remote diagnostics.

[0155] S401: Data Acquisition. Records sub-area ID, trajectory points, module switching time, parameter adjustment records, and monitoring data throughout the entire process.

[0156] S402: Data storage. Store logs in an SQL / NoSQL database, indexed by workpiece ID.

[0157] S403: Report Generation. Generates a PDF / HTML report containing cleaning efficiency curves, residual rate distribution charts, and parameter dynamic curves for quality traceability and analysis.

[0158] S404: Periodic calibration. It is recommended to recalibrate the robot and vision system every 500 hours of operation.

[0159] S405: Remote monitoring. Uploads key operating parameters and alarms to a cloud monitoring platform, allowing maintenance personnel to view data in real time and receive fault alerts. S406: Modular replacement. The quick-change mechanism allows for the replacement of shot blasting or shot peening pumps within 10 minutes, reducing downtime.

[0160] To better illustrate the superior effects of the method in this embodiment, according to Figure 5 The process described herein involves performing a surface treatment operation on a large, complex casting using the method of this invention. The specific steps are as follows: Casting preparation: placing the casting and performing pretreatment. Data acquisition: collecting casting information. Coordinated scheduling and parameter adjustment. Execution of blasting / polishing operation. Evaluation indicators using machine vision. Quality assessment: judging the cleaning quality; if it meets the standard, recording and outputting the results. Otherwise, adjusting and updating parameters until the quality standard is met.

[0161] The workpiece surface has a complex structure, numerous depressions, and dramatic curvature variations. In actual testing, the system, through the intelligent blasting control algorithm of this invention, efficiently and precisely cleans the entire surface, reducing operation time by approximately 22% and increasing the surface treatment pass rate to 98.6%, significantly outperforming traditional manual methods and existing fixed-stroke cleaning solutions.

[0162] Example 2: Embodiment 2 of the present invention provides an intelligent blasting and spraying control system for large and complex castings, such as... Figure 6 As shown, it includes: The workpiece 3D modeling and recognition module is configured to initialize the shot blasting system and shot peening system, place the casting to be cleaned, and perform preprocessing operations for data acquisition.

[0163] The work scheduling module is configured to collect information about the castings to be cleaned, plan the shot blasting operation path based on the collected information, optimize the planned operation path using a genetic algorithm, and schedule the work of the shot blasting system and the shot peening system based on the optimized operation path.

[0164] The work scheduling module includes an intelligent shot blasting path planning module and a work area segmentation and scheduling module. The intelligent shot blasting path planning module collects information about the castings to be cleaned, plans the shot blasting operation path based on the collected information, and optimizes the planned operation path using a genetic algorithm. The work area segmentation and scheduling module schedules the work of the shot blasting system and the shot peening system based on the optimized operation path.

[0165] The feedback adjustment module is configured to calculate the error based on the working conditions of the shot blasting system and the shot peening system, and to adaptively adjust the working parameters according to the error.

[0166] The feedback adjustment module includes an online quality inspection and feedback module and a parameter adaptive optimization module. The online quality inspection and feedback module is used to acquire the operating status of the shot blasting and shot peening systems. The parameter adaptive optimization module is used to adaptively adjust the operating parameters using a reinforcement learning model or fuzzy algorithm.

[0167] It also includes a system control terminal, used to generate and issue control commands.

[0168] Example 3: Embodiment 3 of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the steps of the intelligent blasting control method for large and complex castings as described in Embodiment 1 of the present invention.

[0169] Example 4: Embodiment 4 of the present invention provides a computer device, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the steps of the intelligent blasting and spraying control method for large and complex castings as described in Embodiment 1 of the present invention.

[0170] The steps and methods involved in Examples 2, 3 and 4 above correspond to those in Example 1. For specific implementation details, please refer to the relevant description section of Example 1.

[0171] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)). The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A smart polishing and spraying control method for large and complex castings, characterized in that, Includes the following steps: Initialize the shot blasting and shot peening systems, place the castings to be cleaned, and perform preprocessing operations for data acquisition; Information on the castings to be cleaned is collected, and the shot blasting operation path is planned based on the collected information. The planned operation path is optimized using a genetic algorithm, and the operation scheduling of the shot blasting system and shot peening system is performed based on the optimized operation path. The error is calculated based on the working conditions of the shot blasting system and the shot peening system, and the working parameters are adaptively adjusted according to the error.

2. The intelligent polishing and spraying control method for large and complex castings as described in claim 1, characterized in that, The specific steps for initializing the shot blasting system and shot peening system are as follows: Configure and install the hardware for the initial shot blasting system and shot peening system; Configure and initialize the software for the initial shot blasting system and shot peening system; Check the operating status of the shot blasting system and the shot peening system.

3. The intelligent polishing and spraying control method for large and complex castings as described in claim 1, characterized in that, The specific steps for the preprocessing operation of placing the casting to be cleaned and collecting data are as follows: Model the casting to be cleaned and mesh it; Priority weights are calculated and assigned according to the characteristics of the areas to be operated on the casting; The shot blasting system and shot peening system are performed online for attitude recognition and registration based on the placed castings; The placed castings are divided into work sub-areas.

4. The intelligent polishing and spraying control method for large and complex castings as described in claim 3, characterized in that, Areas with high curvature, deep blind spots, and narrow gaps are given high weight, while flat and easily accessible areas are given low weight.

5. The intelligent polishing and spraying control method for large and complex castings as described in claim 3, characterized in that, The specific steps for dividing the placed castings into work sub-areas are as follows: Divided into multiple sub-regions based on priority weights; The cleanup method is mapped according to the characteristics of each sub-region, and the switching point and cleanup method are recorded.

6. The intelligent polishing and spraying control method for large and complex castings as described in claim 5, characterized in that, The specific steps for planning the spraying operation path based on the collected information are as follows: Construct a sub-region node diagram based on the divided sub-regions, with each node being the center of the grid. Feasible paths are generated based on the combined distance of edge weights and the incident angle penalty.

7. The intelligent polishing and spraying control method for large and complex castings as described in claim 1, characterized in that, The specific steps for calculating the error based on the working conditions of the shot blasting and shot peening systems, and then adaptively adjusting the working parameters according to the error, are as follows: Collect images of the casting surface after cleaning and inspect the cleaning effect; Error calculation is performed based on the test results; Based on fuzzy logic or reinforcement learning methods, the current working parameters are adjusted according to the error.

8. An intelligent polishing and spraying control system for large and complex castings, characterized in that, include: The workpiece 3D modeling and recognition module is configured to initialize the shot blasting system and shot peening system, place the casting to be cleaned, and perform preprocessing operations for data acquisition. The work scheduling module is configured to collect information on the castings to be cleaned, plan the shot blasting operation path based on the collected information, optimize the planned operation path using a genetic algorithm, and schedule the work of the shot blasting system and the shot peening system based on the optimized operation path. The feedback adjustment module is configured to calculate the error based on the working conditions of the shot blasting system and the shot peening system, and to adaptively adjust the working parameters according to the error.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1-7: intelligent blasting control method for large, complex castings.

10. A computer device, characterized in that, A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the intelligent blasting control method for large and complex castings as described in any one of claims 1-7.

Images