Numerical control machine tool machining path optimization method based on artificial intelligence

Abstract

The invention discloses a numerical control machine tool machining path optimization method based on artificial intelligence. The method comprises the steps that S1, a complete input data set needed by an optimization model is generated; s2, according to the complete input data set, constructing a multi-objective optimization model for numerical control machine tool processing path optimization; s3, forming a fuzzy decision basis of the optimization target; s4, determining an initial processing path optimization solution set; s5, generating an improved optimal solution set of the processing path; and S6, checking the improved optimal solution set of the machining path, checking whether the machining path meets the kinematics limitation of the numerical control machine tool, the precision requirement of the machined workpiece and the process constraint of the dynamic machining environment or not, and generating a machining path instruction of the numerical control machine tool after the machining path is checked to be qualified. According to the method, the integration of checking and instruction generation is realized, and the reliability and accuracy of path execution are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of numerical control machine tools, and in particular to a method for optimizing a machining path of a numerical control machine tool based on artificial intelligence. Background Art

[0002] With the rapid development of artificial intelligence technology, CNC machine tool processing in the manufacturing industry is gradually moving towards high precision, high efficiency and intelligence. CNC machine tool processing path optimization is an important link in achieving this goal. It is directly related to processing efficiency, processing quality and machine tool operation stability.

[0003] At present, the planning of CNC machine tool processing paths mainly relies on traditional optimization methods. Traditional optimization methods are usually based on fixed geometric models and single-objective optimization. They generate processing paths by determining the shortest path or the lowest energy consumption. In a static processing environment, relatively ideal results can be obtained. However, in actual applications, the processing environment is often dynamic and complex. For example, dynamic factors such as tool wear, machine tool vibration, and processing temperature changes will significantly affect the actual execution effect of the processing path. Traditional optimization methods are difficult to respond to these dynamic changes in real time, resulting in the inability of the processing path optimization results to adapt to the actual processing scenarios.

[0004] In recent years, artificial intelligence technology has gradually been introduced into the field of CNC machine tool processing. Some studies have tried to use simple genetic algorithms or particle swarm algorithms to improve path optimization effects, but conventional genetic algorithms or particle swarm algorithms still have obvious defects: on the one hand, genetic algorithms or particle swarm algorithms lack the ability to coordinate multi-objective optimization and find it difficult to balance the trade-offs between multiple performance indicators; on the other hand, most algorithms cannot dynamically adjust the path during processing to adapt to changes in the real-time environment. In addition, existing methods usually use independent steps in the evaluation of processing paths and the generation of instructions, and lack a comprehensive verification of path optimization results, which can easily lead to errors or reduced efficiency in the actual execution of processing paths.

[0005] In summary, the existing technology has significant deficiencies in dynamic adaptability, multi-objective collaborative optimization, and the comprehensiveness of path evaluation and instruction generation, and it is difficult to meet the needs of modern manufacturing for efficient, accurate and intelligent path planning. Therefore, a new method is urgently needed to solve the above problems, so as to improve the optimization level and applicability of CNC machine tool processing paths. Summary of the invention

[0006] One purpose of the present invention is to propose a CNC machine tool machining path optimization method based on artificial intelligence. The present invention realizes the integration of verification and instruction generation, and effectively improves the reliability and accuracy of path execution.

[0007] According to an embodiment of the present invention, a method for optimizing a machining path of a CNC machine tool based on artificial intelligence comprises the following steps:

[0008] S1. Collect the geometric data of the workpiece processed by the CNC machine tool, the processing requirements of the workpiece, the performance parameters of the CNC machine tool, and the real-time dynamic parameters of the processing environment, pre-process the collected CNC machine tool processing related data, remove abnormal values ​​and extract key features, and generate a complete input data set required for the optimization model;

[0009] S2. Construct a multi-objective optimization model for CNC machine tool machining path optimization based on the complete input data set. The multi-objective optimization model takes path length, machining time, tool life and energy consumption as optimization objectives, and defines constraints, which include kinematic limitations of the CNC machine tool, precision requirements of the machined workpiece, and process constraints of the dynamic machining environment.

[0010] S3. defining fuzzy membership functions for path length, processing time, tool life and energy consumption respectively, and calculating the fuzzy membership value of the optimization target, wherein the fuzzy membership value is used to quantify the pros and cons of each optimization target, and forms a fuzzy decision-making basis for the optimization target;

[0011] S4. Based on the fuzzy membership function and the optimization model, a fuzzy game optimization method is used to calculate the collaborative solution between the optimization objectives and determine the initial processing path optimization solution set;

[0012] S5. Based on the initial processing path optimization solution set, the bitter fish optimization algorithm is used to perform local search and global expansion in the solution space, and the improved optimization solution set of the processing path is generated by simulating the dynamic update of the position of the bitter fish individual and searching in the area outside the solution set;

[0013] S6. Check the improved optimization solution set of the machining path to see whether the machining path meets the kinematic limitations of the CNC machine tool, the precision requirements of the machined workpiece, and the process constraints of the dynamic machining environment. After the verification is qualified, the machining path instructions of the CNC machine tool are generated.

[0014] Optionally, the S1 includes the following steps:

[0015] S11. Extracting geometric data G of the workpiece through three-dimensional scanning, including length, width, height, edge curvature and surface roughness;

[0016] S12. Collect the processing requirements R of the workpiece, including processing accuracy requirements, surface finish requirements, processing sequence requirements and characteristic parameters of the processing material;

[0017] S13. Record the kinematic performance parameters P of the CNC machine tool, including the maximum feed rate, the maximum spindle speed, the maximum cutting depth, and the current tool wear;

[0018] S14. Real-time monitoring of processing environment parameters E through sensors, including machine tool vibration amplitude, working environment temperature, humidity and processing noise intensity;

[0019] S15. Normalize and remove noise from the collected workpiece geometry data G, processing requirements R, performance parameters P and dynamic environment parameters E. When removing abnormal values, for any parameter set X = {x1, x2, ..., x n The elimination criteria are:

[0020]

[0021] Among them, μ X is the mean of X, σ X is the standard deviation of X, k is the empirical coefficient;

[0022] S16. Combine the preprocessed parameter sets G, R, P, E into the complete input data set D required for the optimization model input :

[0023] D input ={G∪R∪P∪E}.

[0024] Optionally, S2 includes the following steps:

[0025] S21. From the complete input dataset D input Extract workpiece geometry data G, processing requirements R, performance parameters P and dynamic environment parameters E, and analyze the processing state parameters S in real time and dynamically t ={A v ,W t ,T e ,H e}, and build a dynamic weight adjustment mechanism for optimization objectives:

[0026] α i (t) = g i (G,R,P,E,S t );

[0027] Among them, α i (t) is the weight of the optimization target at processing time t, g i It is a weight calculation function based on the input data set and dynamic feedback data. The characteristics of the parameter set are directly used for the target weight priority calculation;

[0028] S22. According to the workpiece geometry data G and processing requirements R in the complete input data set, the complete processing path is divided into n segments, each segment of the path corresponds to a different processing priority, and the optimization model segmentation is defined as follows:

[0029] f i(x k )=α 1,k L(x k )+α 2,k T(x k )+α 3,k C(x k )+α 4,k E(x k );

[0030] Among them, x k is the kth path, α i,k The processing requirements of the segmented path and the dynamic state data are adjusted to make the path optimization meet the local geometric characteristics of the workpiece and the process constraints. The characteristics of each path segment are provided by the complete input data set;

[0031] S23. Using the performance parameter P and the dynamic environment parameter E in the complete input data set, the constraint conditions are divided into global constraints and local constraints;

[0032] The global constraint is based on the overall characteristics of the complete input data set, so that the path global optimization satisfies the total path length L(x) not exceeding the maximum path value L in the process design. max :

[0033]

[0034] The global accuracy error e(x) and surface roughness r(x) meet the accuracy requirements of the workpiece P. r And finish requirements R a :

[0035]

[0036] Local constraints are based on dynamic environment parameters E and tool state W t , adjust the processing speed v(x k ) and cutting depth d(x k ):

[0037] v(x k )≤V max ·(1-βW t ),d(x k )≤D max ·(1-γW t );

[0038] Among them, β and γ are adjustment coefficients affected by the dynamic environment parameter E.

[0039] Optionally, S3 includes the following steps:

[0040] S31. Extract path length L(x), processing time T(x), tool life C(x) and energy consumption E(x), and uniformly define the fuzzy membership function μ of the optimization target i (x), where i∈{L,T,C,E} represents the specific goal:

[0041]

[0042] Among them, x target,i represents the expected value of target i, x max,i represents the maximum value allowed for target i, x is the specific value of path length L(x), machining time T(x), tool life C(x) or energy consumption E(x), μ i (x) is the fuzzy membership value of target i, indicating its quality;

[0043] S32. According to the fuzzy membership values ​​of path length L(x), processing time T(x), tool life C(x) and energy consumption E(x), the comprehensive fuzzy membership value μ(x) is calculated in combination with the dynamic weight adjustment mechanism. The comprehensive fuzzy membership value represents the comprehensive quality of each optimization objective:

[0044] μ(x)=α L μ L (L(x))+α T μ T (T(x))+α C μ C (C(x))+α E μ E (E(x));

[0045] Among them, α L ,α T ,α C ,α E is the dynamic weight coefficient.

[0046] Optionally, S4 includes the following steps:

[0047] S41. According to the comprehensive fuzzy membership value μ(x) and the multi-objective optimization model, the path length L(x), processing time T(x), tool life C(x) and energy consumption E(x) are used as game players to construct a fuzzy game optimization model. The payoff function of each game player is defined as its fuzzy membership value:

[0048] U i (x) = μ i (x),i∈{L,T,C,E};

[0049] Among them, U i (x) is the profit function of the ith objective, μ i(x) is the fuzzy membership value of target i;

[0050] S42. Combine the dynamic weight adjustment mechanism of the multi-objective optimization model to define the strategy space of fuzzy games for:

[0051] S i ={x|L(x),T(x),C(x),E(x) satisfy the constraints of the optimization model};

[0052] Among them, S i is a set of feasible solutions for each objective within its constraints, where the constraints are determined by global and local constraints;

[0053] S43. Using fuzzy game optimization method, calculate the cooperative solution between the optimization objectives, and in the strategy space of each objective The initial solution set is randomly generated, and the profit matrix U of each target in the initial solution set is calculated according to the fuzzy membership value formula:

[0054]

[0055] Each row represents a solution Each column represents the corresponding optimization target benefit;

[0056] Use fuzzy game strategy to solve Nash equilibrium on the payoff matrix U and calculate the equilibrium strategy for each target

[0057]

[0058] Among them, α i is the dynamic weight coefficient in S2;

[0059] S44. Solve fuzzy games collaboratively The output is the initial processing path optimization solution set, and the profit matrix U and strategy distribution data are saved at the same time.

[0060] Optionally, S5 includes the following steps:

[0061] S51. Optimization solution set based on initial machining path As the initial population of the bitter fish optimization algorithm, the population size is N, and the position of each bitter fish individual is represented by X = {x1, x2, ..., x N}, the fitness function of each individual is determined by the comprehensive fuzzy membership value μ(x):

[0062] f(x i )=μ(x i );

[0063] Among them, μ(x i ) is individual xi The comprehensive fuzzy membership value of max , local search radius r l and the global expansion radius r g ;

[0064] S52. For each bitter fish individual x i , in the local search radius r l Generate a new search position x′ based on local constraints and dynamic weight adjustment mechanism i :

[0065] x′ i =x i +λ l ·(x j -x i ));

[0066] Among them, x j The current fitness is higher than x i A random individual, λ l ∈[0,1] is the dynamically adjusted step size coefficient, which is affected by the dynamic environment parameter E and the tool state W t The impact,constraints are controlled by local constraints;

[0067] If the new position x′ i Satisfies the constraints and f(x′ i )>f(x i ), then accept the new position x′ i ; Otherwise keep the original position;

[0068] S53. After each local search, a global expansion search is performed on individuals with lower fitness to simulate their behavior of migrating to a better area in the solution space. The update rule is:

[0069] x″ i =x i +λ g ·(x k -x i );

[0070] Among them, x k is the elite individual with the highest fitness in the current population, λ g ∈[0,1] is the global search step adjustment coefficient, which is dynamically updated by the dynamic weight mechanism. The search range x″ i Satisfy global constraints;

[0071] If x″ i The fitness f(x″) i )>f(x i ), then accept the new position; otherwise keep the original position;

[0072] S54. After each local search and global expansion, the population fitness f(x) is recalculated according to the dynamic weight adjustment mechanism and the fuzzy membership value, and the individuals with the highest fitness are retained as elite individuals;

[0073] S55. Repeat steps S52 to S54 until the maximum number of iterations T is reached. max Or the fitness of the population converges to the threshold ∈, and the solution set with the highest fitness is used as the improved optimization solution set of the processing path.

[0074] Optionally, the S6 includes the following steps:

[0075] S61. For the improved optimization solution set of the machining path, check whether the path meets the kinematic constraints of the CNC machine tool, including whether the feed speed, spindle speed and cutting depth of the path are within the range allowed by the CNC machine tool;

[0076] S62. Check the accuracy requirements of the processing path section by section to ensure that the processing error of the path meets the accuracy standard of the workpiece design. At the same time, evaluate the surface roughness to verify whether the processing quality meets the finish standard required by the design;

[0077] S63. Check the environmental adaptability of the path based on the dynamic processing environment parameters when the machine tool is running, including whether the vibration amplitude of the machine tool, the temperature and humidity of the processing environment meet the process requirements;

[0078] S64. After all verifications are passed, the optimized machining path solution set is converted into machining path instructions that can be directly executed by the CNC machine tool. The instructions include path geometry information, machining process parameters and environmental compensation information.

[0079] The beneficial effects of the present invention are:

[0080] (1) The present invention breaks through the limitation that traditional path planning methods can only handle a single goal and cannot dynamically adapt to complex processing environments by combining a dynamic weight adjustment mechanism with a fuzzy game optimization algorithm. In the process of path optimization, the weights are dynamically adjusted to dynamically balance multiple performance indicators such as path length, processing time, tool life and energy consumption, and multi-objective collaborative optimization is achieved in a fuzzy game framework. Compared with traditional fixed weight or static optimization methods, the present invention can adjust the optimization strategy in real time under complex processing environments, significantly improve the robustness and comprehensive performance of path planning, and ensure the global optimality and adaptability of the processing path in a dynamic environment.

[0081] (2) The present invention introduces the bitter fish optimization algorithm in the process path optimization, and realizes efficient coordination between local search and global expansion by dynamically adjusting the step size and the search radius. The local search is combined with the real-time changes of dynamic environment parameters and tool status to refine the local optimization of the path; the global expansion ensures that the path solution does not fall into the local optimum through the elite retention strategy, while enhancing the diversity of solutions and the global search capability. The bitter fish optimization algorithm is more flexible and can quickly converge to the global optimal solution in the high-dimensional solution space, providing new technical support for the efficient optimization of the process path.

[0082] (3) The present invention systematically verifies whether the path satisfies the kinematic limitations, machining accuracy requirements and process constraints of the dynamic machining environment by comprehensively checking the improved optimization solution set of the machining path. It combines the precise calculation of fuzzy membership values ​​and the dynamic weight mechanism to ensure that the generated machining path instructions can achieve high accuracy and high efficiency in actual execution. By storing the verification data and machining instructions, it provides traceable data support for the iterative improvement of the subsequent path optimization model, realizes the integration of verification and instruction generation, and effectively improves the reliability and accuracy of path execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0084] Figure 1 The present invention provides a flowchart of a method for optimizing machining paths of CNC machines based on artificial intelligence. DETAILED DESCRIPTION

[0085] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0086] refer to Figure 1 , a CNC machine tool machining path optimization method based on artificial intelligence, comprising the following steps:

[0087] S1. Collect the geometric data of the workpiece processed by the CNC machine tool, the processing requirements of the workpiece, the performance parameters of the CNC machine tool, and the real-time dynamic parameters of the processing environment, pre-process the collected CNC machine tool processing related data, remove abnormal values ​​and extract key features, and generate a complete input data set required for the optimization model;

[0088] S2. Construct a multi-objective optimization model for CNC machine tool machining path optimization based on the complete input data set. The multi-objective optimization model takes path length, machining time, tool life and energy consumption as optimization objectives, and defines constraints, which include kinematic limitations of the CNC machine tool, precision requirements of the machined workpiece, and process constraints of the dynamic machining environment.

[0089] S3. define fuzzy membership functions for path length, processing time, tool life and energy consumption respectively, calculate fuzzy membership values ​​of optimization targets, and use fuzzy membership values ​​to quantify the pros and cons of each optimization target, thus forming a fuzzy decision-making basis for the optimization target;

[0090] S4. Based on the fuzzy membership function and optimization model, the fuzzy game optimization method is used to calculate the synergistic solution between the optimization objectives and determine the initial processing path optimization solution set;

[0091] S5. Based on the initial processing path optimization solution set, the bitter fish optimization algorithm is used to perform local search and global expansion in the solution space, and the improved optimization solution set of the processing path is generated by simulating the dynamic update of the position of the bitter fish individual and searching in the area outside the solution set;

[0092] S6. Check the improved optimization solution set of the machining path to see whether the machining path meets the kinematic limitations of the CNC machine tool, the precision requirements of the machined workpiece, and the process constraints of the dynamic machining environment. After the verification is qualified, the machining path instructions of the CNC machine tool are generated.

[0093] In this implementation, S1 includes the following steps:

[0094] S11. Extracting geometric data G of the workpiece through three-dimensional scanning, including length, width, height, edge curvature and surface roughness;

[0095] S12. Collect the processing requirements R of the workpiece, including processing accuracy requirements, surface finish requirements, processing sequence requirements and characteristic parameters of the processing material;

[0096] S13. Record the kinematic performance parameters P of the CNC machine tool, including the maximum feed rate, the maximum spindle speed, the maximum cutting depth, and the current tool wear;

[0097] S14. Real-time monitoring of processing environment parameters E through sensors, including machine tool vibration amplitude, working environment temperature, humidity and processing noise intensity;

[0098] S15. Normalize and remove noise from the collected workpiece geometry data G, processing requirements R, performance parameters P and dynamic environment parameters E. When removing abnormal values, for any parameter set X = {x1, x2, ..., x n The elimination criteria are:

[0099]

[0100] Among them, μ X is the mean of X, σ X is the standard deviation of X, k is the empirical coefficient;

[0101] S16. Combine the preprocessed parameter sets G, R, P, E into the complete input data set D required for the optimization model input :

[0102] D input ={G∪R∪P∪E}.

[0103] In this implementation, S2 includes the following steps:

[0104] S21. From the complete input dataset D input Extract workpiece geometry data G, processing requirements R, performance parameters P and dynamic environment parameters E, and analyze the processing state parameters S in real time and dynamically t ={A v ,W t ,T e ,H e}, and build a dynamic weight adjustment mechanism for optimization objectives:

[0105] α i (t) = g i (G,R,P,E,S t );

[0106] Among them, α i (t) is the weight of the optimization target at processing time t, g i It is a weight calculation function based on the input data set and dynamic feedback data. The characteristics of the parameter set are directly used for the target weight priority calculation;

[0107] S22. According to the workpiece geometry data G and processing requirements R in the complete input data set, the complete processing path is divided into n segments, each segment of the path corresponds to a different processing priority, and the optimization model segmentation is defined as follows:

[0108] f i (x k )=α 1,k L(x k )+α 2,k T(x k )+α 3,k C(x k )+α 4,k E(x k );

[0109] Among them, x kis the kth path, α i,k The processing requirements of the segmented path and the dynamic state data are adjusted to make the path optimization meet the local geometric characteristics of the workpiece and the process constraints. The characteristics of each path segment are provided by the complete input data set;

[0110] S23. Using the performance parameter P and the dynamic environment parameter E in the complete input data set, the constraint conditions are divided into global constraints and local constraints;

[0111] The global constraint is based on the overall characteristics of the complete input data set, so that the path global optimization satisfies the total path length L(x) not exceeding the maximum path value L in the process design. max :

[0112]

[0113] The global accuracy error e(x) and surface roughness r(x) meet the accuracy requirements of the workpiece P. r And finish requirements R a :

[0114]

[0115] Local constraints are based on dynamic environment parameters E and tool state W t , adjust the processing speed v(x k ) and cutting depth d(x k ):

[0116] v(x k )≤V max ·(1-βW t ),d(x k )≤D max ·(1-γW t );

[0117] Among them, β and γ are adjustment coefficients affected by the dynamic environment parameter E.

[0118] In this implementation, S3 includes the following steps:

[0119] S31. Extract path length L(x), processing time T(x), tool life C(x) and energy consumption E(x), and uniformly define the fuzzy membership function μ of the optimization target i (x), where i∈{L,T,C,E} represents the specific goal:

[0120]

[0121] Among them, x target,i represents the expected value of target i, x max,irepresents the maximum value allowed for target i, x is the specific value of path length L(x), machining time T(x), tool life C(x) or energy consumption E(x), μ i (x) is the fuzzy membership value of target i, indicating its quality;

[0122] S32. According to the fuzzy membership values ​​of path length L(x), processing time T(x), tool life C(x) and energy consumption E(x), the comprehensive fuzzy membership value μ(x) is calculated in combination with the dynamic weight adjustment mechanism. The comprehensive fuzzy membership value represents the comprehensive quality of each optimization objective:

[0123] μ(x)=α L μ L (L(x))+α T μ T (T(x))+α C μ C (C(x))+α E μ E (E(x));

[0124] Among them, α L ,α T ,α C ,α E is the dynamic weight coefficient.

[0125] In this implementation, S4 includes the following steps:

[0126] S41. According to the comprehensive fuzzy membership value μ(x) and the multi-objective optimization model, the path length L(x), processing time T(x), tool life C(x) and energy consumption E(x) are used as game players to construct a fuzzy game optimization model. The payoff function of each game player is defined as its fuzzy membership value:

[0127] U i (x) = μ i (x),i∈{L,T,C,E};

[0128] Among them, U i (x) is the profit function of the ith objective, μ i (x) is the fuzzy membership value of target i;

[0129] S42. Combine the dynamic weight adjustment mechanism of the multi-objective optimization model to define the strategy space of fuzzy games for:

[0130] S i ={x|L(x),T(x),C(x),E(x) satisfy the constraints of the optimization model};

[0131] Among them, S iis a set of feasible solutions for each objective within its constraints, where the constraints are determined by global and local constraints;

[0132] S43. Using fuzzy game optimization method, calculate the cooperative solution between the optimization objectives, and in the strategy space of each objective The initial solution set is randomly generated, and the profit matrix U of each target in the initial solution set is calculated according to the fuzzy membership value formula:

[0133]

[0134] Each row represents a solution Each column represents the corresponding optimization target benefit;

[0135] Use fuzzy game strategy to solve Nash equilibrium on the payoff matrix U and calculate the equilibrium strategy for each target

[0136]

[0137] Among them, α i is the dynamic weight coefficient in S2;

[0138] S44. Solve fuzzy games collaboratively The output is the initial processing path optimization solution set, and the profit matrix U and strategy distribution data are saved at the same time.

[0139] In this implementation, S5 includes the following steps:

[0140] S51. Optimization solution set based on initial machining path As the initial population of the bitter fish optimization algorithm, the population size is N, and the position of each bitter fish individual is represented by X = {x1, x2, ..., x N}, the fitness function of each individual is determined by the comprehensive fuzzy membership value μ(x):

[0141] f(x i )=μ(x i );

[0142] Among them, μ(x i ) is individual x i The comprehensive fuzzy membership value of max , local search radius r l and the global expansion radius r g ;

[0143] S52. For each bitter fish individual x i , in the local search radius r l Generate a new search position x′ based on local constraints and dynamic weight adjustment mechanism i :

[0144] x′ i =x i +λ l ·(x j -x i ));

[0145] Among them, x j The current fitness is higher than x i A random individual, λ l ∈[0,1] is the dynamically adjusted step size coefficient, which is affected by the dynamic environment parameter E and the tool state W t The impact,constraints are controlled by local constraints;

[0146] If the new position x′ i Satisfies the constraints and f(x′ i )>f(x′ i ), then accept the new position x′ i ; Otherwise keep the original position;

[0147] S53. After each local search, a global expansion search is performed on individuals with lower fitness to simulate their behavior of migrating to a better area in the solution space. The update rule is:

[0148] x″ i =x i +λ g ·(x k -x i );

[0149] Among them, x k is the elite individual with the highest fitness in the current population, λ g ∈[0,1] is the global search step adjustment coefficient, which is dynamically updated by the dynamic weight mechanism. The search range x″ i Satisfy global constraints;

[0150] If x″ i The fitness f(x″) i )>f(x i ), then accept the new position; otherwise keep the original position;

[0151] S54. After each local search and global expansion, the population fitness f(x) is recalculated according to the dynamic weight adjustment mechanism and the fuzzy membership value, and the individuals with the highest fitness are retained as elite individuals;

[0152] S55. Repeat steps S52 to S54 until the maximum number of iterations T is reached. max Or the fitness of the population converges to the threshold ∈, and the solution set with the highest fitness is used as the improved optimization solution set of the processing path.

[0153] In this implementation, S6 includes the following steps:

[0154] S61. For the improved optimization solution set of the machining path, check whether the path meets the kinematic constraints of the CNC machine tool, including whether the feed speed, spindle speed and cutting depth of the path are within the range allowed by the CNC machine tool;

[0155] S62. Check the accuracy requirements of the processing path section by section to ensure that the processing error of the path meets the accuracy standard of the workpiece design. At the same time, evaluate the surface roughness to verify whether the processing quality meets the finish standard required by the design;

[0156] S63. Check the environmental adaptability of the path based on the dynamic processing environment parameters when the machine tool is running, including whether the vibration amplitude of the machine tool, the temperature and humidity of the processing environment meet the process requirements;

[0157] S64. After all verifications are passed, the optimized machining path solution set is converted into machining path instructions that can be directly executed by the CNC machine tool. The instructions include path geometry information, machining process parameters and environmental compensation information.

[0158] Embodiment 1:

[0159] Example On December 1, 2024, a large aviation manufacturing company began to process a batch of high-precision turbine blades at its CNC machining center in City A. The machine tool model was the XYZ-5000 five-axis CNC machining center. At 9 a.m. that day, the technical team collected geometric data of the blade workpiece through a 3D scanner. The system generated the geometric features of the blade, including a length of 450 mm, a width of 230 mm, and a height of 80 mm. At the same time, it identified the edge curvature and surface roughness of the complex curved surface of the workpiece as 0.03 and 0.9, respectively. Combined with production requirements, the technical team set the machining accuracy error to no more than 0.02 mm, the surface roughness requirement to be within 0.6, and the machining time limit to be completed within 24 hours.

[0160] After the data collection is completed, the system automatically extracts the performance parameters of the CNC machine tool, including the maximum feed speed of 400 mm / s, the maximum spindle speed of 12,000 rpm, and the maximum cutting depth of 5 mm. Subsequently, the system collects the dynamic data of the processing environment in real time through the machine tool sensor: the ambient temperature is 27°C, the humidity is 65%, the machine tool vibration amplitude is 0.3 mm, and the initial wear of the tool is 5%. The above data is preprocessed to form a complete input data set.

[0161] After the system starts the path optimization program, the initial path generation time based on the fuzzy game optimization algorithm is 15 minutes, and the total length of the generated path is 138.2 meters. During the optimization process, the system dynamically adjusts the weight of path optimization in the processing area of ​​the blade's complex curved surface, giving priority to accuracy and tool life. Under the fuzzy game framework, the system calculates the collaborative optimization solution set of path length, processing time, energy consumption and tool wear, and completes the initial path planning.

[0162] At 11 a.m., the processing entered the dynamic optimization stage. The system used the bitter fish optimization algorithm to perform secondary optimization on the initial path. During the local search, the system found that the processing time of the tool in the curved area increased by 8% than expected, and the vibration amplitude rose to 0.45 mm in a short time. The system adjusted the path optimization parameters in real time and recalculated the local path feed speed of the tool, reducing the impact of the vibration amplitude on the accuracy. In the global expansion stage, the system redistributed the processing load on specific path segments and adjusted the spindle speed from 12,000 rpm to 11,000 rpm, extending the service life of the tool.

[0163] By 3 p.m., the system checked the generated optimized path. The inspection results showed that the feed speed, spindle speed and cutting depth of the processing path all met the kinematic limitations of the machine tool. The processing error was 0.016 mm and the surface roughness was 0.58, both of which met the production requirements. After the path verification was completed, the system automatically generated the processing path instructions and sent them to the CNC machine tool.

[0164] During the actual processing, the system recorded the processing efficiency data. The processing was completed at 7 am the next day. The actual processing time was 22.5 hours, which was 2.5 hours less than the traditional method. The tool life was extended from 12 hours of the traditional method to 16 hours, and the energy consumption was reduced from the original 64 kWh to 57 kWh. After the processing was completed, the system stored the verification data and path optimization records, and generated an analysis report of the optimized path.

[0165] Through the examples, the method of the present invention fully demonstrates the dynamic adaptability and multi-objective optimization capabilities in complex processing scenarios, which not only improves processing efficiency and quality, but also significantly reduces production costs and energy consumption, providing reliable support for enterprises to achieve intelligent production.

[0166] The present invention breaks through the limitations of traditional path planning methods that can only handle a single goal and cannot dynamically adapt to complex processing environments by combining a dynamic weight adjustment mechanism with a fuzzy game optimization algorithm. In the process of path optimization, the weights are dynamically adjusted to dynamically balance multiple performance indicators such as path length, processing time, tool life and energy consumption, and multi-objective collaborative optimization is achieved in a fuzzy game framework. Compared with traditional fixed weight or static optimization methods, the present invention can adjust the optimization strategy in real time under complex processing environments, significantly improve the robustness and comprehensive performance of path planning, and ensure the global optimality and adaptability of the processing path in a dynamic environment.

[0167] The present invention introduces a bitter fish optimization algorithm in machining path optimization, and realizes efficient collaboration between local search and global expansion by dynamically adjusting the step size and the search radius. The local search is combined with the real-time changes of dynamic environmental parameters and tool status to refine the local optimization of the path; the global expansion ensures that the path solution does not fall into the local optimum through an elite retention strategy, while enhancing the diversity of solutions and the global search capability. The bitter fish optimization algorithm is more flexible and can quickly converge to the global optimal solution in a high-dimensional solution space, providing new technical support for the efficient optimization of machining paths.

[0168] The present invention systematically verifies whether the path satisfies the kinematic limitations, machining accuracy requirements and process constraints of the dynamic machining environment by comprehensively checking the improved optimized solution set of the machining path, and combines the precise calculation of fuzzy membership values ​​and the dynamic weight mechanism to ensure that the generated machining path instructions can achieve high accuracy and high efficiency in actual execution. By storing the verification data and machining instructions, traceable data support is provided for the iterative improvement of the subsequent path optimization model, realizing the integration of verification and instruction generation, and effectively improving the reliability and accuracy of path execution.

[0169] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A CNC machine tool machining path optimization method based on artificial intelligence, characterized in that: The steps include: S1. Collect the geometric data of the workpiece processed by the CNC machine tool, the processing requirements of the workpiece, the performance parameters of the CNC machine tool, and the real-time dynamic parameters of the processing environment, pre-process the collected CNC machine tool processing related data, remove abnormal values ​​and extract key features, and generate a complete input data set required for the optimization model; S2. Construct a multi-objective optimization model for CNC machine tool machining path optimization based on the complete input data set. The multi-objective optimization model takes path length, machining time, tool life and energy consumption as optimization objectives, and defines constraints, which include kinematic limitations of the CNC machine tool, precision requirements of the machined workpiece, and process constraints of the dynamic machining environment. S3. defining fuzzy membership functions for path length, processing time, tool life and energy consumption respectively, and calculating the fuzzy membership value of the optimization target, wherein the fuzzy membership value is used to quantify the pros and cons of each optimization target, and forms a fuzzy decision-making basis for the optimization target; S4. Based on the fuzzy membership function and the optimization model, a fuzzy game optimization method is used to calculate the collaborative solution between the optimization objectives and determine the initial processing path optimization solution set; S5. Based on the initial processing path optimization solution set, the bitter fish optimization algorithm is used to perform local search and global expansion in the solution space, and the improved optimization solution set of the processing path is generated by simulating the dynamic update of the position of the bitter fish individual and searching in the area outside the solution set; S6. Check the improved optimization solution set of the machining path to see whether the machining path meets the kinematic limitations of the CNC machine tool, the precision requirements of the machined workpiece, and the process constraints of the dynamic machining environment. After the verification is qualified, the machining path instructions of the CNC machine tool are generated.

2. The method for optimizing machining paths of CNC machine tools based on artificial intelligence according to claim 1, characterized in that: The S1 comprises the following steps: S11. Extracting geometric data G of the workpiece through three-dimensional scanning, including length, width, height, edge curvature and surface roughness; S12. Collect the processing requirements R of the workpiece, including processing accuracy requirements, surface finish requirements, processing sequence requirements and characteristic parameters of the processing material; S13. Record the kinematic performance parameters P of the CNC machine tool, including the maximum feed rate, the maximum spindle speed, the maximum cutting depth, and the current tool wear; S14. Real-time monitoring of processing environment parameters E through sensors, including machine tool vibration amplitude, working environment temperature, humidity and processing noise intensity; S15. Normalize and remove noise from the collected workpiece geometry data G, processing requirements R, performance parameters P and dynamic environment parameters E. When removing abnormal values, for any parameter set X = {x1, x2, ..., x n The elimination criteria are: Among them, μ X is the mean of X, σ X is the standard deviation of X, k is the empirical coefficient; S16. Combine the preprocessed parameter sets G, R, P, E into the complete input data set D required for the optimization model input : D input ={G∪R∪P∪E}。 3. The method for optimizing machining paths of CNC machine tools based on artificial intelligence according to claim 1, characterized in that: The S2 comprises the following steps: S21. From the complete input dataset D input Extract workpiece geometry data G, processing requirements R, performance parameters P and dynamic environment parameters E, and analyze the processing state parameters S in real time and dynamically. t ={A v ,W t ,T e ,H e }, and build a dynamic weight adjustment mechanism for optimization objectives: α i (t)=g i (G,R,P,E,S t ); Among them, α i (t) is the weight of the optimization target at processing time t, g i It is a weight calculation function based on the input data set and dynamic feedback data. The characteristics of the parameter set are directly used for the target weight priority calculation; S22. According to the workpiece geometry data G and processing requirements R in the complete input data set, the complete processing path is divided into n segments, each segment of the path corresponds to a different processing priority, and the optimization model segmentation is defined as follows: f i (x k )=a 1,k L(x k )+a 2,k T(x k )+a 3,k C(x k )+a 4,k E(x k ); Among them, x k is the kth path, α i,k The processing requirements of the segmented path and the dynamic state data are adjusted to make the path optimization meet the local geometric characteristics of the workpiece and the process constraints. The characteristics of each path segment are provided by the complete input data set; S23. Using the performance parameter P and the dynamic environment parameter E in the complete input data set, the constraint conditions are divided into global constraints and local constraints; The global constraint is based on the overall characteristics of the complete input data set, so that the path global optimization satisfies the total path length L(x) not exceeding the maximum path value L in the process design. max : The global accuracy error e(x) and surface roughness r(x) meet the accuracy requirements of the workpiece P. r And finish requirements R a : Local constraints are based on dynamic environment parameters E and tool state W t , adjust the processing speed v(x k ) and cutting depth d(x k ): v(x k )≤V max ·(1-βW t ),d(x k )≤D max ·(1-γW t ); Among them, β and γ are adjustment coefficients affected by the dynamic environment parameter E.

4. The method for optimizing machining paths of CNC machine tools based on artificial intelligence according to claim 1, characterized in that: The S3 comprises the following steps: S31. Extract path length L(x), processing time T(x), tool life C(x) and energy consumption E(x), and uniformly define the fuzzy membership function μ of the optimization target i (x), where i∈{L,T,C,E} represents the specific goal: Among them, x target,i represents the expected value of target i, x max,i represents the maximum value allowed for target i, x is the specific value of path length L(x), machining time T(x), tool life C(x) or energy consumption E(x), μ i (x) is the fuzzy membership value of target i, indicating its quality; S32. According to the fuzzy membership values ​​of path length L(x), processing time T(x), tool life C(x) and energy consumption E(x), the comprehensive fuzzy membership value μ(x) is calculated in combination with the dynamic weight adjustment mechanism. The comprehensive fuzzy membership value represents the comprehensive quality of each optimization objective: μ(x)=a L m L (L(x))+α T m T (T(x))+α C m C (C(x))+α E m E (E(x)); Among them, α L ,α T ,α C ,α E is the dynamic weight coefficient.

5. The method for optimizing machining paths of CNC machine tools based on artificial intelligence according to claim 1, characterized in that: The S4 comprises the following steps: S41. According to the comprehensive fuzzy membership value μ(x) and the multi-objective optimization model, the path length L(x), processing time T(x), tool life C(x) and energy consumption E(x) are used as game players to construct a fuzzy game optimization model. The payoff function of each game player is defined as its fuzzy membership value: U i (x)=μ i (x),i∈{L,T,C,E}; Among them, U i (x) is the profit function of the i-th objective, μ i (x) is the fuzzy membership value of target i; S42. Combine the dynamic weight adjustment mechanism of the multi-objective optimization model to define the strategy space of fuzzy games for: S i ={x|L(x),T(x),C(x),E(x) satisfy the constraints of the optimization model}; Among them, S i is a set of feasible solutions for each objective within its constraints, where the constraints are determined by global and local constraints; S43. Using fuzzy game optimization method, calculate the cooperative solution between the optimization objectives, and in the strategy space of each objective The initial solution set is randomly generated, and the profit matrix U of each target in the initial solution set is calculated according to the fuzzy membership value formula: Each row represents a solution Each column represents the corresponding optimization target benefit; Use fuzzy game strategy to solve Nash equilibrium on the payoff matrix U and calculate the equilibrium strategy for each target Among them, α i is the dynamic weight coefficient in S2; S44. Solve fuzzy games collaboratively The output is the initial processing path optimization solution set, and the profit matrix U and strategy distribution data are saved at the same time.

6. The method for optimizing machining paths of CNC machine tools based on artificial intelligence according to claim 1, characterized in that: The S5 comprises the following steps: S51. Optimization solution set based on initial machining path As the initial population of the bitter fish optimization algorithm, the population size is N, and the position of each bitter fish individual is represented by X = {x1, x2, ..., x N }, the fitness function of each individual is determined by the comprehensive fuzzy membership value μ(x): f(x i )=μ(x i ); Among them, μ(x i ) is individual x i The comprehensive fuzzy membership value of max , local search radius r l and the global expansion radius r g ; S52. For each bitter fish individual x i , in the local search radius r l Generate a new search position x′ based on local constraints and dynamic weight adjustment mechanism i : x′ i =x i +λ l ·(x j -x i )); Among them, x j The current fitness is higher than x i A random individual, λ l ∈[0,1] is the dynamically adjusted step size coefficient, which is affected by the dynamic environment parameter E and the tool state W t The impact,constraints are controlled by local constraints; If the new position x′ i Satisfies the constraints and f(x′ i )>f(x i ), then accept the new position x′ i ; Otherwise keep the original position; S53. After each local search, a global expansion search is performed on individuals with lower fitness to simulate their behavior of migrating to a better area in the solution space. The update rule is: x″ i =x i +λ g ·(x k -x i ); Among them, x k is the elite individual with the highest fitness in the current population, λ g ∈[0,1] is the global search step adjustment coefficient, which is dynamically updated by the dynamic weight mechanism. The search range x″ i Satisfy global constraints; If x″ i The fitness f(x″) i )>f(x i ), then accept the new position; otherwise keep the original position; S54. After each local search and global expansion, the population fitness f(x) is recalculated according to the dynamic weight adjustment mechanism and the fuzzy membership value, and the individuals with the highest fitness are retained as elite individuals; S55. Repeat steps S52 to S54 until the maximum number of iterations T is reached. max Or the fitness of the population converges to the threshold ∈, and the solution set with the highest fitness is used as the improved optimization solution set of the processing path.

7. The method for optimizing machining paths of CNC machine tools based on artificial intelligence according to claim 1, characterized in that: The S6 comprises the following steps: S61. For the improved optimization solution set of the machining path, check whether the path meets the kinematic constraints of the CNC machine tool, including whether the feed speed, spindle speed and cutting depth of the path are within the range allowed by the CNC machine tool; S62. Check the accuracy requirements of the processing path section by section to ensure that the processing error of the path meets the accuracy standard of the workpiece design. At the same time, evaluate the surface roughness to verify whether the processing quality meets the finish standard required by the design; S63. Check the environmental adaptability of the path based on the dynamic processing environment parameters when the machine tool is running, including whether the vibration amplitude of the machine tool, the temperature and humidity of the processing environment meet the process requirements; S64. After all verifications are passed, the optimized machining path solution set is converted into machining path instructions that can be directly executed by the CNC machine tool. The instructions include path geometry information, machining process parameters and environmental compensation information.

Images