Complex surface machining region decomposition and process parameter optimization method for multiple processes

Abstract

The present invention discloses a method for complex surface processing area decomposition and process parameter optimization for multiple processes, and belongs to the technical field of robot processing of complex surface workpieces. The method first performs parameterization and discretization processing on the surface model of the complex surface, decomposes the complex surface by clustering partitions and defining boundaries to obtain multiple units, and then calculates the process operation complexity in combination with the geometric complexity of the workpiece surface and related process parameters, and adjusts the complex surface decomposition according to the process operation complexity of each unit, analyzes the influence of process parameters on operation efficiency and optimizes the process parameters in each unit, and finally completes the process adaptive unit decomposition of the complex surface, and obtains the most efficient process parameter combination, which lays a good foundation for improving the quality and efficiency of complex surface processing.

Description

Technical Field

[0001] The invention belongs to the technical field of robot processing of complex curved surface workpieces, and in particular relates to a method for multi-process oriented complex curved surface processing region decomposition and process parameter optimization. Background Art

[0002] At present, in the fields of aerospace, medical equipment, new energy, etc., high-precision and complex curved surface components such as aircraft skins, car bodies, and fan blades have been widely used, and various industries have also put forward more stringent requirements for their manufacturing and processing technologies. With the improvement of the level of automation and intelligence in industrial production, complex surface processing technology has developed from the initial manual operation to more efficient CNC lathe processing and industrial robot processing. Industrial robots are widely used in process tasks such as spraying, grinding, welding, drilling and laser processing due to their advantages such as high degrees of freedom and easy programming and control. However, due to the variable structural characteristics of complex surfaces and the complex process parameters involved, industrial robot processing still has the problem of not being able to balance processing efficiency and accuracy. Therefore, most researchers partition complex surfaces to reduce the difficulty of processing. However, the existing surface decomposition methods are often based on the geometric properties of the processing object, lacking consideration of different process tasks, and the optimization analysis of process parameters is also mainly focused on a single process, which restricts the improvement of processing efficiency and accuracy.

[0003] Chinese patent CN107562013A discloses a method for planning variable tool machining trajectories for surfaces with equal residual heights. The patent partitions the surface according to the geometric features of the surface and the machining requirements to determine the boundaries of adjacent machining areas, and initially plans machining trajectories with equal residual heights for different machining areas. By analyzing the geometric features at the boundary, the radius of curvature of the short-range line and the tool connection line spacing along the line spacing direction are calculated, and then the tool contact points and tool position points corresponding to any point on the boundary of the adjacent machining areas are solved to form a tool position point boundary. By adjusting all tool position points, the errors caused by re-planning are averaged, the tool connection marks are reduced, and the machining quality is improved. However, this method does not fully consider the impact of different process tasks on the partition results, as well as the detailed adjustment of the process parameters in each partition, and has limitations in the application of complex surface multi-process machining.

[0004] Chinese patent CN112666890A discloses a curved workpiece machining trajectory planning method, which divides the curved workpiece to be machined into multiple machining areas, and optimizes the machining trajectory using orthogonal decomposition or minimum angle methods according to the size of the curvature. The path planning efficiency is greatly improved, and the collision problem is avoided by rationally connecting the moving points. However, this method does not take into account the factors affecting the machining quality, and does not consider the impact of changes in the curvature characteristics of complex surfaces on the selection of process parameters, which has certain limitations.

[0005] Therefore, it is necessary to study a method of complex surface machining area decomposition and process parameter optimization for multiple processes to meet the requirements of high-precision and high-efficiency process machining. Summary of the invention

[0006] The present invention discloses a method for complex surface machining area decomposition and process parameter optimization for multiple processes. The surface is partitioned according to the complex surface geometric characteristics and process requirements. Through comprehensive analysis of the surface geometric characteristics and process parameters, a quantitative estimation of the complexity of the complex surface process operation is obtained to achieve more reasonable surface decomposition and process parameter selection, thereby effectively solving at least one technical problem involved in the background technology.

[0007] To achieve the above object, the technical solution of the present invention is:

[0008] A method for complex surface machining region decomposition and process parameter optimization for multiple processes, comprising the following steps:

[0009] Step S1, modeling a complex surface to obtain a surface model;

[0010] Step S2, using parameter lines to discretize the complex surface, extracting discrete points on the surface and judging their types, to obtain multiple discrete point subsets with different characteristics;

[0011] Step S3, using the fuzzy c-means clustering algorithm to calculate the center of each discrete point subset, clustering the surface discrete points with similar characteristics by minimizing the objective function, connecting the centers of adjacent discrete point subsets in the area to be processed, and dividing the boundary based on the perpendicular bisector to obtain multiple units, completing the initial decomposition of the surface;

[0012] Step S4, based on the multiple units obtained by the initial decomposition of the surface, the complexity of the process operation is calculated by comprehensively considering the geometric features and process parameters;

[0013] Step S5, traversing all units obtained by the initial decomposition of the surface, performing secondary subdivision on the units whose process complexity exceeds the threshold based on the K-means clustering method, using the surface roughness of the complex surface workpiece and the grinding and polishing efficiency as the evaluation parameters of the workpiece performance, and constructing a complex workpiece surface roughness prediction model for predicting the workpiece surface roughness based on the known grinding and polishing process parameters and process complexity and using an artificial neural network;

[0014] Step S6, training the constructed complex workpiece surface roughness prediction model with grinding and polishing sample data;

[0015] Step S7, based on the trained complex workpiece surface roughness prediction model and combined with the grinding and polishing efficiency, a multi-objective optimization mathematical model is established, the process parameters are optimized through a genetic algorithm and the multi-objective optimization mathematical model is solved to obtain the optimal process parameter combination for each unit.

[0016] Optionally, in step S1, the surface S(u, v) of the surface model is expressed by the following formula:

[0017] ;

[0018] In the formula, For along u Direction i The number of p of B Spline function, For along v Direction j The number of q of B Spline function, Corresponding to the control point The weight factor of is the point at index ( i,j ), m is the number of rows of control points, and n is the number of columns of control points.

[0019] Optionally, in step S3, the objective function is minimized It is expressed by the following formula:

[0020] ;

[0021] In the formula, b is the fuzzy weighted index, b ≥1; U is the fuzzy partitioning matrix; V is the cluster center matrix; a is the number of data sets; c is the number of cluster centers; d rk For the k The samples and r The Euclidean distance between cluster centers; For the r Cluster centers k The fuzzy membership of samples and meets the following three constraints:

[0022] ;

[0023] ;

[0024] .

[0025] Optionally, in step S4, the process operation complexity C is calculated by the following formula:

[0026] ;

[0027] In the formula, w 1 and w 2 are the weights of geometric and task complexity in process complexity respectively; α s is the range of the mean curvature of the surface, indicating the s The geometric complexity of each unit; β s It is a process parameter that has a strong correlation with the processing quality; k s It is the ratio of the actual surface area to the projected surface area.

[0028] Optionally, in step S5, the arithmetic mean deviation of the peak-to-valley profile is used as an evaluation parameter for the surface processing quality of the complex curved workpiece, wherein the surface roughness Ra of the complex curved workpiece is expressed by the following formula:

[0029] ;

[0030] In the formula, is the sampling length, f(x) represents the roughness curve, dx For variables x Infinitesimal increments of

[0031] The workpiece surface roughness change rate is used as the evaluation parameter for grinding and polishing efficiency, where the workpiece surface roughness change rate E f It is expressed by the following formula:

[0032] ;

[0033] In the formula, A is the polishing surface area, T For the polishing time, is the surface roughness variation.

[0034] Optionally, in step S5, the known grinding and polishing process parameters include the initial roughness of the workpiece surface, the grinding and polishing pressure, the grinding tool rotation speed, the feed speed, the number of grinding and polishing times, the grit of the grinding and polishing sandpaper, and the hardness of the workpiece material.

[0035] Optionally, in step S5, the complex workpiece surface roughness prediction model includes an input layer, a hidden layer and an output layer, wherein the process parameters input by the input layer include process operation complexity, initial roughness of the workpiece surface, grinding and polishing pressure, grinding tool speed, feed speed, grinding and polishing times, sandpaper mesh size and workpiece material hardness; the parameters output by the output layer are the surface roughness of the workpiece after processing.

[0036] Optionally, the activation functions of the input layer and output layer of the complex workpiece surface roughness prediction model are hyperbolic tangent S-type function and linear function respectively, and the LM algorithm is selected as the training function, and the mean square error is used as the error control function; the remaining parameters of the model are set as: the maximum number of training times is 5000, the learning rate is 0.01, the learning rate growth ratio is 1, the decline ratio is 0.7, and the momentum factor is 0.02.

[0037] Optionally, in step S6, before training the constructed complex workpiece surface roughness prediction model with grinding and polishing sample data, the input grinding and polishing sample data needs to be preprocessed, specifically including:

[0038] The input grinding and polishing sample data is normalized by linear transformation. The input range of the processed data is [0.1, 0.9]. The linear transformation expression is:

[0039] ;

[0040] In the formula, is the data value before processing; is the processed data value; and are the maximum and minimum values ​​in the grinding and polishing sample data;

[0041] The anti-normalized expression of the output is:

[0042] ;

[0043] Where: is the output data after denormalization; is the output data before denormalization; and are the maximum and minimum values ​​in the sample output data set, respectively.

[0044] Optionally, in step S7, the process parameters are optimized by a genetic algorithm and a multi-objective optimization mathematical model is solved, specifically including:

[0045] Step S71, select optimization parameters, and set the initial roughness of the workpiece surface , Workpiece material hardness H BAssuming the complexity of the process is known, the process parameters to be optimized are: Grinding and polishing pressure P , grinding tool speed n s , Feed speed v a , grinding and polishing times N And grinding sandpaper mesh G ;

[0046] Step S72, determining the objective function and constraint conditions of the multi-objective optimization mathematical model to minimize the surface roughness of the complex curved workpiece R a and the highest grinding and polishing efficiency E f Optimize

[0047] ;

[0048] In the formula, 5kPa≤ P ≤20kPa; 120mm / min≤ v a ≤240mm / min; 800r / min≤ n s ≤1200r / min; 3 times≤ N ≤12 times; 0.1≤ G ≤0.9; R 1 is the processing radius;

[0049] Step S73, construct a fitness function, which is expressed as follows:

[0050] ;

[0051] In the formula, Q is a constant, w 1 and w 2 The objective functions are f 1 and f 2 The weight coefficient of

[0052] Step S74, solving a multi-objective optimization mathematical model, specifically includes:

[0053] Binary coding is used to initialize and create a population, and the surface roughness prediction model of complex workpieces and the grinding and polishing efficiency are calculated. Then the fitness value of each individual is calculated. When the fitness value reaches the set end condition, the individual with the largest fitness value is taken as the optimal result and decoded and output. Otherwise, the genetic operation continues until the end condition is met.

[0054] The beneficial effects of the present invention are as follows:

[0055] In the application context of multi-process high-precision robot machining, the present invention aims to address the problems in which, during the robot machining of complex curved workpieces, the use of unified conservative parameters for the entire surface leads to severely limited operating efficiency, and the existing surface decomposition method cannot flexibly adapt to changes in process tasks, resulting in uneven distribution of machining errors. By designing a process operation complexity evaluation method based on the combination of complex surface geometric features and process planning related parameters, the present invention completes the initial decomposition of complex surfaces based on fuzzy C-means clustering and the regional subdivision based on the complexity of sub-surface process operations. On this basis, the law of change of process efficiency with process parameters is analyzed, and finally the process parameter optimization of partitioned machining of complex curved workpieces is realized, the optimal process parameters under different operation complexities are determined, the regional adaptability of each sub-surface process planning in complex operation tasks is improved, the robot machining efficiency is optimized, and technical support is provided for the high-efficiency machining of complex curved workpieces. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0057] Figure 1 It is a flow chart of a method for complex surface machining area decomposition and process parameter optimization for multiple processes of the present invention;

[0058] Figure 2 It is a schematic diagram of a curved surface model of a complex curved surface of the present invention;

[0059] Figure 3 It is a schematic diagram of the surface model after discretization processing of the present invention;

[0060] Figure 4 This is a schematic diagram of clustering of discrete points on a surface according to the present invention;

[0061] Figure 5 This is a schematic diagram of the boundary division of discrete points on a curved surface according to the present invention;

[0062] Figure 6 It is a schematic diagram of calculating the surface roughness of a workpiece according to the present invention;

[0063] Figure 7 It is a schematic diagram of the structure of the complex workpiece surface roughness prediction model of the present invention;

[0064] Figure 8 The present invention is a flow chart for solving the multi-objective optimization mathematical model. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present invention will be described clearly and completely below in combination with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0067] In addition, in the present invention, descriptions such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0068] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0069] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0070] When using industrial robots to process complex surfaces, due to the large changes in model curvature, traditional global processing methods or partitioned processing methods based only on surface geometric features usually cannot fully consider the impact of changes in complex workpiece surface curvature and changes in multiple process tasks on robot processing efficiency. That is, there is no task planning that adapts to the process, resulting in low efficiency and high energy consumption in robot processing. In view of this situation, in order to synergistically improve the robot's processing accuracy and efficiency, the present invention provides a method for regional decomposition and process parameter optimization of complex surface processing for multiple processes. Taking the bunny model and grinding and polishing process in the Stanford 3D scanning model dataset as an example, with the help of MATLAB software, the overall process is as follows: Figure 1 As shown, the specific steps include:

[0071] Step S1, modeling a complex surface to obtain a surface model;

[0072] Since the surface curvature of the surface model often has problems such as mutation, the NURBS parameterized surface S (u, v) is established to accurately describe the surface characteristics of the surface model, such as Figure 2 As shown, the surface S(u, v) of the surface model is expressed as follows:

[0073] ;

[0074] In the formula, For along u Direction i The number of p of B Spline function, For along v Direction j The number of q of B Spline function, Corresponding to the control point The weight factor of is the point at index ( i,j ), m is the number of rows of control points, and n is the number of columns of control points.

[0075] Since the geometric characteristics of a surface can be reflected by the first basic form and the second basic form of the surface, if we take any point on the surface S(u, v), its direction is expressed as (du:dv), then the first basic form I of the surface S(u, v) is:

[0076] ;

[0077] In the formula, E, F, and G are the first-class basic quantities of the surface, and the expressions are as follows:

[0078] ;

[0079] It follows that the unit normal vector of any point on the surface S(u, v) is:

[0080] ;

[0081] The second basic form II of the surface S(u, v) is:

[0082] ;

[0083] In the formula, L, M, and N are the second-class basic quantities of the surface, and the expressions are as follows:

[0084] ;

[0085] One of the principal curvature directions of the surface S(u, v) is ∂S(u, v) = S(u, v)∂u + S(u, v)∂v, then the calculation of (∂u,∂v) satisfies:

[0086] ;

[0087] In the formula, λ is the characteristic value.

[0088] The above equation has a non-zero solution, so the determinant of the coefficient matrix is ​​zero, that is:

[0089] ;

[0090] Expand to get:

[0091] ;

[0092] From the perspective of differential geometry, the above equation has two solutions (principal curvatures), denoted by k max and k min , let K = k max k min and H = (k max +k min ) / 2, K is the Gaussian curvature, H is the mean curvature, and the expressions are:

[0093] ;

[0094] In the above equation, E , F , G , L , M and N Determines the Gaussian curvature K and mean curvature H The size ofK and H is an intrinsic property of the surface and is independent of direction. The degree of local surface curvature can be determined by K The positive or negative value of the local surface can be reflected by H is represented by positive or negative.

[0095] Step S2, using parameter lines to discretize the complex surface, extracting discrete points on the surface and judging their types, to obtain multiple discrete point subsets with different characteristics;

[0096] The complex surface is discretized and decomposed using parameter lines u and v of appropriate density, wherein the complexity of the surface, the required calculation accuracy and the calculation time affect the sparsity of the parameter lines. In this specific embodiment, according to the size of the actual surface, 50u-direction and 50v-direction parameter lines are selected to discretize the complex surface, such as Figure 3 When the parameter lines are too sparse, the shape of the surface cannot be truly reflected with fewer grids, thus affecting the subsequent division of sub-regions; however, when the density of the parameter lines exceeds a certain level, the number of sub-regions will not change with the grid.

[0097] Step S3, using the fuzzy c-means clustering algorithm to calculate the center of each discrete point subset, clustering the surface discrete points with similar characteristics by minimizing the objective function (see Figure 4 As shown in Figure 2), connect the centers of adjacent discrete point subsets in the area to be processed, and divide the boundary based on the perpendicular bisector to obtain multiple units to complete the initial decomposition of the surface. Figure 5 As shown;

[0098] Among them, the minimization objective function It is expressed by the following formula:

[0099] ;

[0100] In the formula, b is the fuzzy weighted index, b ≥1; U is the fuzzy partitioning matrix; V is the cluster center matrix; a is the number of data sets; c is the number of cluster centers; d rk For the k The samples and r The Euclidean distance between cluster centers; For the r Cluster centers k The fuzzy membership of samples and meets the following three constraints:

[0101] ;

[0102] ;

[0103] .

[0104] Step S4, based on the multiple units obtained by the initial decomposition of the surface, the complexity of the process operation is calculated by comprehensively considering the geometric features and process parameters;

[0105] It should be noted that the process parameters include process parameters for various tasks, such as processing tool size, cutting method, spraying pressure, etc.

[0106] The process complexity C is calculated by the following formula:

[0107] ;

[0108] In the formula, w 1 and w 2 are the weights of geometric and task complexity in process complexity respectively; α s is the range of the mean curvature of the surface, indicating the s The geometric complexity of the unit is:

[0109] ;

[0110] In the formula, It is s The maximum mean curvature of the units; It is s The minimum mean curvature of the elements.

[0111] β s It is a process parameter that has a strong correlation with the processing quality. Taking the grinding and polishing process as an example, β s Indicates processing s The tool diameter effect coefficient of each unit is expressed as:

[0112] ;

[0113] In the formula, D is the tool diameter, r min is the minimum curvature radius of the surface. The larger the tool diameter coefficient, the greater the ratio of the surface diameter to the minimum curvature radius, and the greater the complexity of complex surface machining.

[0114] Step S5, traversing all units obtained by the initial decomposition of the surface, performing secondary subdivision on the units whose process complexity exceeds the threshold based on the K-means clustering method, using the surface roughness of the complex surface workpiece and the grinding and polishing efficiency as the evaluation parameters of the workpiece performance, and constructing a complex workpiece surface roughness prediction model for predicting the workpiece surface roughness based on the known grinding and polishing process parameters and process complexity and using an artificial neural network;

[0115] It should be noted that a reasonable process operation complexity threshold should be set according to specific process requirements and precision requirements to balance processing quality and production efficiency.

[0116] Specifically, the efficiency of robot grinding and polishing operations is related to process parameters and the complexity of industrial operations in the sub-area. For the automated grinding and polishing of complex curved workpieces, the reasonable and effective selection and control of grinding and polishing process parameters are the key to ensuring high-quality and efficient processing. The main process parameters of the robot grinding and polishing system include grinding and polishing pressure, abrasive tool speed, feed speed, abrasive tool size, and grinding and polishing times. In a specific embodiment of the present invention, the arithmetic mean deviation of the peak-to-valley profile is used as the evaluation parameter of the surface roughness Ra of complex curved workpieces. It represents the arithmetic mean of the absolute value of the distance from each point on the measured micro-convex profile to its profile midline within a certain selected sample length, which can vividly reflect the microscopic geometric shape error characteristics of the workpiece surface, such as Figure 6 As shown, the surface roughness Ra of the complex curved workpiece is expressed by the following formula:

[0117] ;

[0118] In the formula, is the sampling length, f(x) represents the roughness curve, dx For variables x Infinitesimal increments of

[0119] The surface roughness change rate of the workpiece is used to evaluate the grinding and polishing efficiency. The surface roughness change rate of the workpiece refers to the change rate of the workpiece surface roughness for a given grinding and polishing surface area. A , during the polishing time T The ratio of the product of the change in workpiece surface roughness and surface area to time, among which the grinding and polishing efficiency E f It is expressed by the following formula:

[0120] ;

[0121] In the formula, is the surface roughness variation.

[0122] In actual processing, the efficiency of grinding and polishing is often related to time, that is, it is defined as the shortest processing time and the highest efficiency for grinding and polishing workpieces of the same area while meeting the surface processing quality.

[0123] The known grinding and polishing process parameters include the initial roughness of the workpiece surface, grinding and polishing pressure, grinding tool speed, feed speed, grinding and polishing times, grinding and polishing sandpaper mesh size and workpiece material hardness.

[0124] The performance of robot grinding and polishing operations is related to process parameters and the complexity of sub-area process operations. For the automated grinding and polishing of complex curved workpieces, the reasonable and effective selection and control of grinding and polishing process parameters are the key to ensuring high-quality and efficient processing. The main process parameters of the robot grinding and polishing system include grinding and polishing pressure, abrasive tool speed, feed speed, abrasive tool size and grinding and polishing times.

[0125] Combination Figure 7 As shown, the complex workpiece surface roughness prediction model is an 8-input 1-output structure, including an input layer, a hidden layer and an output layer, wherein the process parameters input by the input layer include process operation complexity, initial roughness of the workpiece surface, grinding and polishing pressure, grinding tool speed, feed speed, grinding and polishing times, sandpaper mesh number and workpiece material hardness; the parameters output by the output layer are the surface roughness of the workpiece after processing.

[0126] The activation functions of the input layer and output layer of the complex workpiece surface roughness prediction model are hyperbolic tangent S-type function (tansig) and linear function (purelin), respectively, and the LM algorithm (trainlm) is selected as the training function, and the mean square error (MSE) is used as the error control function; the remaining parameters of the model are set as follows: the maximum number of training times is 5000, the learning rate is 0.01, the learning rate growth ratio is 1, the decline ratio is 0.7, and the momentum factor is 0.02.

[0127] The activation functions of the input layer and the output layer are the hyperbolic tangent sigmoid function (tansig) and the linear function (purelin), respectively. The LM algorithm (trainlm) is selected as the training function, and the mean square error (MSE) is used as the error control function. The remaining parameters of the model are set as follows: the maximum number of training times is 5000, the learning rate is 0.01, the learning rate growth ratio is 1, the decline ratio is 0.7, and the momentum factor is 0.02.

[0128] Step S6, training the constructed complex workpiece surface roughness prediction model with grinding and polishing sample data;

[0129] The training ends when the deviation between the actual output and the expected output of the complex workpiece surface roughness prediction model reaches the set value.

[0130] It should be noted that this model can be used to predict the surface roughness of the workpiece under the premise of known machining process parameters.

[0131] In addition, before training the constructed complex workpiece surface roughness prediction model with grinding and polishing sample data, the input grinding and polishing sample data needs to be preprocessed, including:

[0132] The input grinding and polishing sample data is normalized by linear transformation. The input range of the processed data is [0.1, 0.9]. The linear transformation expression is:

[0133] ;

[0134] In the formula, is the data value before processing; is the processed data value; and are the maximum and minimum values ​​in the grinding and polishing sample data;

[0135] The anti-normalized expression of the output is:

[0136] ;

[0137] Where: is the output data after denormalization; is the output data before denormalization; and are the maximum and minimum values ​​in the sample output data set, respectively.

[0138] Step S7, based on the trained complex workpiece surface roughness prediction model and combined with the grinding and polishing efficiency, a multi-objective optimization mathematical model is established, the process parameters are optimized through a genetic algorithm and the multi-objective optimization mathematical model is solved to obtain the optimal process parameter combination for each unit.

[0139] The process parameters are optimized through genetic algorithms and the multi-objective optimization mathematical model is solved, including:

[0140] Step S71, select optimization parameters, and set the initial roughness of the workpiece surface , Workpiece material hardness H B Assuming the complexity of the process is known, the process parameters to be optimized are: Grinding and polishing pressure P , grinding tool speed n s , Feed speed v a , grinding and polishing times N And grinding sandpaper mesh G ;

[0141] Step S72, determining the objective function and constraint conditions of the multi-objective optimization mathematical model to minimize the surface roughness of the complex curved workpiece R a and the highest grinding and polishing efficiency E f Optimize

[0142] ;

[0143] In the formula, 5kPa≤ P ≤20kPa; 120mm / min≤ v a ≤240mm / min; 800r / min≤ n s ≤1200r / min; 3 times≤ N ≤12 times; 0.1≤ G ≤0.9; R 1 is the processing radius;

[0144] Step S73, construct a fitness function, which is expressed as follows:

[0145] ;

[0146] In the formula, Q is a constant, w 1 and w 2 The objective functions are f 1 and f 2 The weight coefficient of

[0147] The weight coefficient method is used to construct the fitness function, that is, the dual-objective optimization problem of comprehensive processing quality and efficiency is transformed into a single-objective optimization problem, thereby solving the problem of multi-objective optimization conflicts. In addition, in order to eliminate the difference in units and orders of magnitude between the two objective functions when calculating the fitness value, it is necessary to first process the data of the two objective function values ​​and then calculate their fitness values.

[0148] Step S74, solving the multi-objective optimization mathematical model, see Figure 8 As shown, specifically including:

[0149] Binary coding is used to initialize and create a population, and the surface roughness prediction model of complex workpieces and the grinding and polishing efficiency are calculated. Then the fitness value of each individual is calculated. When the fitness value reaches the set end condition, the individual with the largest fitness value is taken as the optimal result and decoded and output. Otherwise, the genetic operation continues until the end condition is met.

[0150] In this way, the adaptive unit decomposition and process parameter optimization of complex surfaces are completed, laying a good foundation for the subsequent robot's operation point and coverage path planning.

[0151] It should be further explained that the genetic algorithm is used to solve the multi-objective optimization mathematical model, the end condition is set as the evolutionary generation of the population, and the two optimization goals are the minimum surface roughness and the highest grinding and polishing efficiency.

[0152] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A method for complex surface machining region decomposition and process parameter optimization for multiple processes, characterized in that: The following steps are involved: Step S1, modeling a complex surface to obtain a surface model; Step S2, using parameter lines to discretize the complex surface, extracting discrete points on the surface and judging their types, to obtain multiple discrete point subsets with different characteristics; Step S3, using the fuzzy c-means clustering algorithm to calculate the center of each discrete point subset, clustering the surface discrete points with similar characteristics by minimizing the objective function, connecting the centers of adjacent discrete point subsets in the area to be processed, and dividing the boundary based on the perpendicular bisector to obtain multiple units, completing the initial decomposition of the surface; Step S4, based on the multiple units obtained by the initial decomposition of the surface, the complexity of the process operation is calculated by comprehensively considering the geometric features and process parameters; Step S5, traversing all units obtained by the initial decomposition of the surface, performing secondary subdivision on the units whose process complexity exceeds the threshold based on the K-means clustering method, using the surface roughness of the complex surface workpiece and the grinding and polishing efficiency as the evaluation parameters of the workpiece performance, and constructing a complex workpiece surface roughness prediction model for predicting the workpiece surface roughness based on the known grinding and polishing process parameters and process complexity and using an artificial neural network; Step S6, training the constructed complex workpiece surface roughness prediction model with grinding and polishing sample data; Step S7, based on the trained complex workpiece surface roughness prediction model and combined with the grinding and polishing efficiency, a multi-objective optimization mathematical model is established, the process parameters are optimized through a genetic algorithm and the multi-objective optimization mathematical model is solved to obtain the optimal process parameter combination for each unit.

2. The method for complex surface processing region decomposition and process parameter optimization for multiple processes according to claim 1 is characterized in that: In step S1, the surface S(u, v) of the surface model is expressed as follows: ； In the formula, For along u Direction i The number of p of B Spline function, For along v Direction j The number of q of B Spline function, Corresponding to the control point The weight factor of is the point at index ( i,j ), m is the number of rows of control points, and n is the number of columns of control points.

3. The method for complex surface machining region decomposition and process parameter optimization for multiple processes according to claim 2 is characterized in that: In step S3, the objective function is minimized It is expressed by the following formula: ； In the formula, b is the fuzzy weighted index, b ≥1; U is the fuzzy partitioning matrix; V is the cluster center matrix; a is the number of data sets; c is the number of cluster centers; d rk For the k The samples and r The Euclidean distance between cluster centers; For the r Cluster centers k The fuzzy membership of samples and meets the following three constraints: ； ； 。 4. The method for complex surface machining region decomposition and process parameter optimization for multiple processes according to claim 3 is characterized in that: In step S4, the process complexity C Calculated by the following formula: ； In the formula, w 1 and w 2 are the weights of geometric and task complexity in process complexity respectively; α s is the range of the mean curvature of the surface, indicating the s The geometric complexity of each unit; β s It is a process parameter that has a strong correlation with the processing quality; k s It is the ratio of the actual surface area to the projected surface area.

5. The method for complex surface processing region decomposition and process parameter optimization for multiple processes according to claim 4 is characterized in that: In step S5, the arithmetic mean deviation of the peak-to-valley profile is used as the evaluation parameter of the surface roughness Ra of the complex curved surface workpiece, wherein the surface roughness Ra of the complex curved surface workpiece is expressed by the following formula: ； In the formula, is the sampling length, f(x) represents the roughness curve, dx For variables x Infinitesimal increments of The surface roughness change rate of the workpiece is used as the evaluation parameter of the grinding and polishing efficiency. E f It is expressed by the following formula: ； In the formula, A is the polishing surface area, T For the polishing time, is the surface roughness variation.

6. The method for complex surface processing region decomposition and process parameter optimization for multiple processes according to claim 5 is characterized in that: In step S5, the known grinding and polishing process parameters include the initial roughness of the workpiece surface, the grinding and polishing pressure, the grinding tool rotation speed, the feed speed, the number of grinding and polishing times, the grinding and polishing sandpaper mesh and the hardness of the workpiece material.

7. The method for complex surface machining region decomposition and process parameter optimization for multiple processes according to claim 6, characterized in that: In step S5, the complex workpiece surface roughness prediction model includes an input layer, a hidden layer and an output layer, wherein the process parameters input by the input layer include process operation complexity, initial roughness of the workpiece surface, grinding and polishing pressure, grinding tool speed, feed speed, grinding and polishing times, sandpaper mesh number and workpiece material hardness; the parameters output by the output layer are the surface roughness of the workpiece after processing.

8. The method for complex surface machining region decomposition and process parameter optimization for multiple processes according to claim 7, characterized in that: The activation functions of the input layer and output layer of the complex workpiece surface roughness prediction model are hyperbolic tangent S-type function and linear function respectively, and the LM algorithm is selected as the training function, and the mean square error is used as the error control function; the remaining parameters of the model are set as: the maximum number of training times is 5000, the learning rate is 0.01, the learning rate growth ratio is 1, the decline ratio is 0.7, and the momentum factor is 0.

02.

9. The method for multi-process complex surface machining region decomposition and process parameter optimization according to claim 8, characterized in that: In step S6, before training the constructed complex workpiece surface roughness prediction model with grinding and polishing sample data, the input grinding and polishing sample data needs to be preprocessed, specifically including: The input grinding and polishing sample data is normalized by linear transformation. The input range of the processed data is [0.1, 0.9]. The linear transformation expression is: ； In the formula, is the data value before processing; is the processed data value; and are the maximum and minimum values ​​in the grinding and polishing sample data; The anti-normalized expression of the output is: ； Where: is the output data after denormalization; is the output data before denormalization; and are the maximum and minimum values ​​in the sample output data set, respectively.

10. The method for complex surface machining region decomposition and process parameter optimization for multiple processes according to claim 9, characterized in that: In step S7, the process parameters are optimized by genetic algorithm and a multi-objective optimization mathematical model is solved, which specifically includes: Step S71, select optimization parameters, and set the initial roughness of the workpiece surface , Workpiece material hardness H B Assuming the complexity of the process is known, the process parameters to be optimized are: Grinding and polishing pressure P , grinding tool speed n s , Feed speed v a , grinding and polishing times N And grinding sandpaper mesh G ; Step S72, determining the objective function and constraint conditions of the multi-objective optimization mathematical model to minimize the surface roughness of the complex curved workpiece R a and the highest grinding and polishing efficiency E f Optimize ； In the formula, 5kPa≤ P ≤20kPa; 120mm / min≤ v a ≤240mm / min; 800r / min≤ n s ≤1200 r / min; 3 times≤ N ≤12 times; 0.1≤ G ≤0.9; R 1 is the processing radius; Step S73, construct a fitness function, which is expressed as follows: ； In the formula, Q is a constant, w 1 and w 2 The objective functions are f 1 and f 2 The weight coefficient of Step S74, solving a multi-objective optimization mathematical model, specifically includes: Binary coding is used to initialize and create a population, and the surface roughness prediction model of complex workpieces and the grinding and polishing efficiency are calculated. Then the fitness value of each individual is calculated. When the fitness value reaches the set end condition, the individual with the largest fitness value is taken as the optimal result and decoded and output. Otherwise, the genetic operation continues until the end condition is met.

Images