High-efficiency automatic control platform for multi-axis machining of parts

Abstract

The invention discloses a high-efficiency automatic control platform for multi-axis machining of parts, relates to the technical field of machining automation, and aims to solve the problems that the evaluation of the machining difficulty of the parts always depends on artificial experience or simple rules, so that the subjectivity is high, the evaluation is not accurate enough, and the production cost is high in the existing manufacturing process which always depends on a large amount of artificial intervention and experience. And a full-automatic and intelligent decision support system is lacked. According to different processing difficulty degrees of parts caused by different strength and hardness of raw materials and different processing step number and tolerance requirements, the parts which are difficult to process are processed step by step according to the classification of the difficulty degrees of the parts, and the system can automatically select proper strategies and algorithms according to actual processing requirements, so that the processing efficiency is improved. The machining efficiency and quality are improved, human intervention is reduced, the production efficiency is improved, the system can conduct self-adaptive adjustment according to real-time data in the machining process, and the precision and flexibility of the machining process are improved.

Description

Technical Field

[0001] The present invention relates to the field of machining automation technology, and more specifically, to a high-performance component multi-axis machining automation control platform. Background Art

[0002] With the continuous development of modern manufacturing, the requirements for machining accuracy, efficiency and automation of parts are getting higher and higher, especially in the fields of aerospace, automobile, precision instrument, mold manufacturing, etc. The design of parts is becoming more and more complex, the size requirements are becoming more and more precise, and higher requirements are put forward for the automation and intelligence level of the machining process. For this reason, multi-axis machining technology for high-efficiency parts has emerged and gradually become an important part of the advanced manufacturing field. Traditional manual programming and operation methods are inefficient, require a lot of manual intervention, are prone to errors, and are difficult to cope with complex machining requirements. There is a lack of intelligent scheduling and real-time feedback mechanisms, and the utilization rate of machine tools may be low, resulting in a waste of production resources.

[0003] The prior art has the following deficiencies:

[0004] In the existing technology, the evaluation of part processing difficulty often relies on manual experience or simple rules, which may lead to strong subjectivity and inaccurate evaluation. The selection of appropriate optimization algorithms often relies on experience or trial and error, and lacks intelligent adaptability for specific problems. In complex processing processes with multiple steps and multiple factors, existing technologies often have difficulty dealing with the complex interactions between steps and their impact on the final processing results, and are unable to model and optimize these complex processing processes to improve the controllability and optimization potential of the entire processing process. In general, the existing manufacturing process often relies on a lot of manual intervention and experience, and lacks a fully automated and intelligent decision support system.

[0005] In view of the above problems, the present invention proposes a solution. Summary of the invention

[0006] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a high-performance component multi-axis machining automation control platform to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] High-performance component multi-axis machining automation control platform, including: information acquisition module, data processing module, step-by-step preprocessing module and algorithm selection module, signal connection between modules;

[0009] Information acquisition module: obtains the strength and hardness of the raw materials of the parts, the number of processing steps of the parts, and the required part tolerances;

[0010] Data processing module: Calculate the raw material characteristic coefficient of the part according to the strength and hardness of the raw material of the part, and calculate the processing difficulty coefficient according to the number of processing steps of the part and the required part tolerance, and then weight the comprehensive processing difficulty coefficient to determine whether the part is easy to process;

[0011] Step-by-step preprocessing module: If it is not easy to process, the part steps are divided into steps, and the time series is used to calculate the correlation between the steps and the size of the recorded state space;

[0012] Algorithm selection module: Determines whether it is more appropriate to use the policy gradient method or the Q-Learning exploration learning algorithm based on the correlation between steps and the size of the state space, selects whether the next step needs to be changed, and selects the optimal step to optimize the process.

[0013] In a preferred embodiment, the information acquisition module includes the following contents:

[0014] Record the strength and hardness of raw materials by referring to material manuals or standards;

[0015] Arrange reasonable steps according to the functions and requirements of the parts by process engineers, or simulate the processing of the parts through CAD and CAE software, and record the number of processing steps of the parts;

[0016] The part tolerance shall be based on the requirements of the part. If there is no clear tolerance requirement, refer to the general standards, check the relevant industry standards or technical requirements, and record the part tolerance.

[0017] In a preferred embodiment, the data processing module includes the following contents:

[0018] The material characteristic coefficient is calculated based on the hardness and strength of the raw material of the part, using the following formula: K m =α·H+β·σ, where: K m is the material characteristic coefficient, H is the hardness of the material, σ is the strength of the material, α and β are the weight coefficients of hardness and strength to processing difficulty;

[0019] Set a weighting factor to include the number of processing steps and tolerance requirements into the criteria for judging whether a part is easy to process. The comprehensive processing difficulty coefficient is calculated using the following formula: K process =γ·N+δ·T, where: K process is the processing difficulty coefficient, N is the number of processing steps, T is the tolerance requirement of the part, γ and δ are the weighting coefficients of the number of processing steps and tolerance requirements;

[0020] Combining the material property coefficient and the processing difficulty coefficient, the comprehensive processing difficulty coefficient is expressed by the following formula: K total=λ·K m +(1-λ)·K process , where: K total is the comprehensive processing difficulty coefficient, and λ is a weighting coefficient representing the relative importance between material properties and processing difficulty;

[0021] Set a threshold to judge the processing difficulty, set a threshold K threshold , judge according to the following rules: If K total ≤K threshold , the parts are easy to process and the processing difficulty is relatively low K total >K threshold , the parts are not easy to process and the processing difficulty is high.

[0022] In a preferred embodiment, the step-by-step preprocessing module includes the following contents:

[0023] For parts that are not easy to process, the processing is broken down into more detailed steps, recorded as S1, S2, S3...S i ..., where i is the number of the step, S i is the i-th step;

[0024] In the dependency matrix method of the processing process, a matrix is ​​constructed to represent the dependency relationship between the steps, where each matrix element A ij Indicates step S i and step S j If A ij is a non-zero element, it means step S j For step S i There is a direct dependency relationship. By counting the number of non-zero elements in the matrix, the size of the direct dependency of each step is further derived and normalized;

[0025] For each step, the following information is recorded: operation type and operation parameters. The size of the state space is the number of combinations of all these operation parameters and state information. The operation type and parameters of each step can be selected independently, so the size of the state space of the step is the product of the number of operation types and the number of operation parameters, which is expressed as: state space = operation type × operation parameter.

[0026] In a preferred embodiment, the algorithm selection module includes the following contents:

[0027] After receiving the correlation between steps and the state space, the algorithm selection module defines the correlation between steps and the state space as input variables and divides them into different fuzzy sets;

[0028] The exploratory learning algorithm is defined as the output variable, which is divided into fuzzy sets;

[0029] Formulate fuzzy rules to describe the connections between steps and the impact of state space definition on exploration learning algorithms;

[0030] Carry out fuzzy reasoning based on fuzzy rules and determine the exploration learning algorithm;

[0031] Using the determined exploration learning algorithm, between two step nodes, before proceeding to the next step, the probability of continuing to use the next step and the probability of using another step are calculated, and the probability of normally proceeding to the next step is marked as t. At this node, there is another 1-t probability of choosing another step. The probabilities of the two steps are compared to decide whether to proceed to the next step normally or choose another step. The decision is made based on the value of t. If t is higher, choose to continue using the next step, otherwise choose another step.

[0032] The technical effects and advantages of the high-performance multi-axis machining automation control platform for parts of the present invention are as follows:

[0033] The present invention takes into account the different strengths and hardness of raw materials and the different numbers of processing steps and tolerance requirements that lead to different processing difficulties for each part. According to the difficulty classification of parts, parts that are not easy to process are processed step by step, and the reinforcement learning algorithm to be used is determined according to the size of the time series correlation and the size of the state space. The processing steps and the learning algorithm used can be intelligently adjusted. In this way, the system can automatically select appropriate strategies and algorithms according to actual processing needs, improve processing efficiency and quality, further enhance the automation and intelligence of the processing process, reduce human intervention, and improve production efficiency. The system can make adaptive adjustments based on real-time data during the processing process to improve the accuracy and flexibility of the processing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the structure of the high-performance component multi-axis machining automation control platform of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in 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.

[0036] Example 1

[0037] The invention discloses a high-performance component multi-axis machining automation control platform, comprising: an information acquisition module, a data processing module, a step-by-step preprocessing module and an algorithm selection module, and the modules are signal-connected.

[0038] Information acquisition module: obtains the strength and hardness of the raw materials of the parts, the number of processing steps of the parts, and the required part tolerances;

[0039] Data processing module: Calculate the raw material characteristic coefficient of the part according to the strength and hardness of the raw material of the part, and calculate the processing difficulty coefficient according to the number of processing steps of the part and the required part tolerance, and then weight the comprehensive processing difficulty coefficient to determine whether the part is easy to process;

[0040] Step-by-step preprocessing module: If it is not easy to process, the part steps are divided into steps, and the time series is used to calculate the correlation between the steps and the size of the recorded state space;

[0041] Algorithm selection module: Determines whether it is more appropriate to use the policy gradient method or the Q-Learning exploration learning algorithm based on the correlation between steps and the size of the state space, selects whether the next step needs to be changed, and selects the optimal step to optimize the process.

[0042] The information acquisition module is used to obtain the strength and hardness of the raw materials of the parts, the number of processing steps of the parts, and the required part tolerances. The specific process is as follows:

[0043] By referring to material manuals or standards, such as the "Steel Manual", "Mechanical Engineering Manual" or national standards, you can understand the strength and hardness of raw materials. Common engineering materials have detailed physical and mechanical properties data, and the strength and hardness of the raw materials are recorded.

[0044] Strength: usually includes yield strength, tensile strength, fatigue strength, etc., which refers to the ability of the material surface to resist being pressed or scratched by external objects. Simply put, the higher the hardness, the less likely the material is to be scratched or deformed. Hardness is usually an indicator used to describe the surface characteristics of a material, which is mainly related to the microstructure of the material.

[0045] Hardness: The hardness value affects the wear resistance and processability of parts. It refers to the ability of the material to resist damage or deformation under external forces. It indicates the maximum load that the material can withstand without damage or fracture, and reflects the deformation or fracture limit of the material under external forces. The greater the strength of the raw material, it usually means that the material has a higher resistance to external forces, pressure or other physical loads.

[0046] The processing steps of parts are generally clearly stated in the process route or process card. Process engineers will make reasonable step arrangements based on the functions and requirements of the parts, or simulate the processing of the parts through CAD and CAE software. The software can provide recommended processing steps based on the design and processing technology of the parts, and record the number of processing steps for the resulting parts.

[0047] The part tolerance shall be based on the requirements of the part. If there is no clear tolerance requirement, or if a general standard is required, consult the relevant industry standards or technical requirements, refer to industry standards, such as: ISO 2768: General tolerance standard, applicable to most non-precision parts, GB / T 1804-2000: China's general tolerance standard, and record the part tolerance.

[0048] The data processing module calculates the raw material characteristic coefficient of the part according to the strength and hardness of the raw material of the part, and calculates the processing difficulty coefficient according to the number of processing steps of the part and the required part tolerance, and makes a weighted judgment on whether the part is easy to process. The specific process is as follows:

[0049] The material characteristic coefficient is usually calculated based on the hardness and strength of the raw material of the part, which will directly affect the difficulty of part processing. The material characteristic coefficient can be calculated using the following formula: K m =α·H+β·σ, where: K m is the material characteristic coefficient, H is the hardness of the material, σ is the strength of the material, α and β are the weight coefficients of hardness and strength on processing difficulty. These weight coefficients can be determined through experience or experiments. Usually, hardness has a greater impact on processing difficulty, so hardness may be given a greater weight.

[0050] Set a weighting factor to include the number of processing steps and tolerance requirements into the criteria for judging whether a part is easy to process. The comprehensive processing difficulty coefficient is calculated using the following formula: K process =γ·N+δ·T, where: K process is the processing difficulty coefficient, N is the number of processing steps, T is the tolerance requirement of the part, γ and δ are the weighted coefficients of the number of processing steps and tolerance requirements, which are usually set through experience or process data.

[0051] Combining the material property coefficient and the processing difficulty coefficient, the comprehensive processing difficulty coefficient is expressed by the following formula: K total =λ·K m +(1-λ)·K process , where: K total is the comprehensive processing difficulty coefficient, and λ is a weighting coefficient representing the relative importance between material properties and processing difficulty.

[0052] Set a threshold to judge the processing difficulty. Once the comprehensive processing difficulty coefficient K is obtained,total , a threshold can be set to determine whether a part is "easy to process" or "not easy to process". The specific judgment can be made by comparing it with empirical data or industry standards and setting a threshold K threshold , judge according to the following rules: If K total ≤K threshold , the parts are easy to process and the processing difficulty is relatively low K total >K threshold , the parts are not easy to process and the processing difficulty is high.

[0053] For ease of understanding, the following example is given: Assume the following data: material hardness H = 150HB, material strength σ = 600MPa, number of processing steps N = 10; tolerance requirement T = 0.05mm, set weighting coefficients α = 0.6, β = 0.4, γ = 0.5, δ = 0.5, λ = 0.7.

[0054] Calculate material characteristic coefficient: K m =α·H+β·σ=330, calculate the processing difficulty coefficient: K process =γ·N+δ·T=5.025, calculate the comprehensive processing difficulty coefficient: K total =λ·K m +(1-λ)·K process =232.5075, assuming the threshold is K threshold =200, to judge whether the parts are easy to process: K total >K threshold Therefore, it is judged that the part is not easy to process.

[0055] Step-by-step preprocessing module: If it is not easy to process, the part steps are divided into steps, and the time series is used to calculate the correlation between the steps and the size of the recorded state space. The specific process is as follows:

[0056] For parts that are not easy to process, the processing process should be broken down into more detailed steps. Each processing step may include some sub-steps or operations, such as rough processing, fine processing, heat treatment, inspection and finishing, surface treatment, and each large processing step can be broken down into small operable steps, which helps to more clearly understand the complexity of the processing process and the relationship between each step, recorded as S1, S2, S3...S i ..., where i is the number of the step, S i is the ith step.

[0057] In the dependency matrix method of the processing process, we construct a matrix to represent the dependency relationship between the steps, where each matrix element A ij Indicates step S i and step Sj If A ij is a non-zero element, it means step S j For step S i There is a direct dependency relationship. The matrix can be used to analyze and calculate the dependencies between steps. By counting the number of non-zero elements in the matrix, the direct dependency size of each step (that is, the degree of their influence on other steps) can be further derived and normalized so that the final dependency size is unified to [0,1] for subsequent analysis.

[0058] For example: Let's take a three-order matrix consisting of three steps for a simple explanation. The matrix A is: Among them A 12 =1, is a non-zero element, which means that step 2 is dependent on step 1, A 23 =0, which is a zero element, proves that there is no dependency between step 3 and step 2, and there is only one non-zero element in the matrix.

[0059] It should be noted that if you want to quantify the strength of the dependency, you can use not only 0 and 1 in the matrix, but also other values ​​(such as 0.1, 0.5, 0.8, etc.) to represent different degrees of dependency. The calculation is to count the number of non-zero elements. In addition to direct dependency, you can also calculate compound dependency (indirect dependency) through path analysis. This is usually achieved through matrix multiplication, for example, by calculating the power of the matrix (A 2 ,A 3 ,…) can be used to derive the impact of indirect dependencies, which can be set by technical personnel and will not be described in detail here.

[0060] The size of the state space usually depends on the optional operations of each step and the possible states of each step. For each step, the following information is recorded: operation type (such as cutting, welding, etc.) and operation parameters (such as cutting depth, angle, etc.). The size of the state space is the number of combinations of all these operation parameters and state information. The operation type and parameters of each step can be selected independently, so the size of the state space of the step is the product of the number of operation types and the number of operation parameters, which can be expressed as: state space = operation type × operation parameter.

[0061] For example: suppose there is a manufacturing process, in which different operation types and parameters of each operation can be selected in the steps: There are 3 types of operation types: cutting, welding, and grinding. Operation parameters Each operation has 5 parameter options. For example: Cutting operation: cutting depth (1, 2, 3, 4, 5) and angle (45°, 90°, 135°, 180°). Welding operation: welding temperature (low, medium, high) and duration (short, medium, long). Grinding operation: grinding speed (slow, medium, fast) and force (light, medium, heavy). For each operation, the number of combinations of operation type and operation parameters can be calculated: Cutting: There are 5 depth options and 4 angle options, so the number of combinations of cutting operations is 5×4=20. Welding: There are 3 temperature options and 3 duration options, so the number of combinations of welding operations is 3×3=9. Grinding: There are 3 speed options and 3 force options, so the number of combinations of grinding operations is 3×3=9.

[0062] Therefore, the state space size for each step is the number of operation types multiplied by the number of combinations of corresponding operation parameters. If there are 3 types of operations (cutting, welding, grinding), their state space size is: 3×(20+9+9)=114. Therefore, the state space size of this step is 114.

[0063] The algorithm selection module determines whether it is more appropriate to use the policy gradient method or the Q-Learning exploration learning algorithm based on the correlation between the steps and the size of the state space to select whether the next step needs to be changed and select the optimal step optimization process. The specific process is as follows:

[0064] Q-Learning is a value-based reinforcement learning algorithm that selects the optimal strategy by estimating the value function Q(s,a) of each state-action pair. The core idea of ​​Q-Learning is to update the Q value through the Bellman equation to gradually approach the optimal strategy. Q-Learning is a value-based offline algorithm. When interacting with the environment, it does not require explicit strategy learning, but indirectly improves the strategy by updating the Q value. By updating the Q value of each state-action pair, the optimal action is gradually found.

[0065] The policy gradient method maximizes the cumulative reward by optimizing the policy itself. Unlike Q-Learning, the policy gradient method directly parameterizes the policy and selects the optimal action through gradient updates.

[0066] After receiving the correlation between steps and the state space, the algorithm selection module defines the correlation between steps and the state space as input variables and divides them into different fuzzy sets.

[0067] For example, "Low", "Medium", "High" for the dependencies between steps, and "Small", "Medium", "Large" for the state space.

[0068] Define the exploratory learning algorithm as the output variable, which is divided into fuzzy sets, for example, "Q-Learning", "Gradient" for the exploratory learning algorithm.

[0069] Formulate a set of fuzzy rules to describe the impact of different input variables on output variables. The definition of rules can be based on professional knowledge or obtained through data analysis and experiments. For example:

[0070] Label the association between steps as G, the state space as Z, and the exploration learning algorithm as Algorithm, then we can define:

[0071] Rule 1:IF(G is Low)AND(Z is Small)THEN(Algorithm is Q-Learning)

[0072] Rule 2:IF(G is High)AND(Z is Large)THEN(Algorithm is Gradient) ...

[0074] Perform fuzzy reasoning based on fuzzy rules and determine the solution for exploring learning algorithms.

[0075] It should be noted that the division of fuzzy sets can be adjusted according to actual conditions. For example, although this embodiment takes three fuzzy sets as examples, the correlation between steps, the state space, and the exploration learning algorithm can actually be divided into more than three sets to facilitate better precise adjustment according to different situations.

[0076] Furthermore, for the judgment of the correlation between steps and the high, medium and low of the state space, thresholds can be set according to actual conditions. For example, when the correlation between steps exceeds 0.7, it is marked as "High", and when the state space is higher than 100, it is marked as "Large", and so on. I will not go into details here.

[0077] The determined exploration learning algorithm can be used to calculate the probability of continuing to use the next step or using another step before proceeding to the next step between two step nodes. The probability of proceeding to the next step normally is marked as t, and there is another 1-t probability of selecting another step at this node. Comparing the probabilities of the two steps determines whether to proceed to the next step normally or select another step.

[0078] Assuming the step nodes S1 and S2, which step do you need to choose at this node? According to the current node status, calculate the probability t of choosing the next step S1 and the probability 1-t of choosing step S2. Make a decision based on the value of t. If t is high, choose step S1, otherwise choose step S2. Choose the current optimal action, that is, when a slight change occurs during work, the part processing effect will be evaluated at the time point of the exploration probability. If the part processing effect is good, make adjustments based on the slight change difference.

[0079] It should be noted that the above modules applied to the high-performance multi-axis machining automation control platform for parts can be classified according to the difficulty of part machining, calculate the correlation between the steps of parts that are not easy to machine and the size of the state space, and automatically select the most appropriate learning algorithm. Through intelligent algorithm selection, the platform can continuously optimize the machining process based on real-time data and machining feedback, and ultimately improve the efficiency and quality of the entire parts machining process, which can reduce unnecessary time waste in the machining process, reduce the error rate, and thus reduce production costs. In addition, based on the optimized machining path and strategy, faster production cycles can be achieved and production efficiency can be improved, which enables the platform to make the best decisions when facing parts of different types and complexities, thereby improving the intelligence level of the entire production system.

[0080] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0081] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0082] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application of the technical solution and the invention constraints. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0083] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0084] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0085] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. High-performance multi-axis machining automation control platform for parts, characterized by: include: Information acquisition module, data processing module, step-by-step preprocessing module and algorithm selection module, and signal connections between the modules; Information acquisition module: obtains the strength and hardness of the raw materials of the parts, the number of processing steps of the parts, and the required part tolerances; Data processing module: Calculate the raw material characteristic coefficient of the part according to the strength and hardness of the raw material of the part, and calculate the processing difficulty coefficient according to the number of processing steps of the part and the required part tolerance, and then weight the comprehensive processing difficulty coefficient to determine whether the part is easy to process; Step-by-step preprocessing module: If it is not easy to process, the part steps are divided into steps, and the time series is used to calculate the correlation between the steps and the size of the recorded state space; Algorithm selection module: Determines whether it is more appropriate to use the policy gradient method or the Q-Learning exploration learning algorithm based on the correlation between steps and the size of the state space, selects whether the next step needs to be changed, and selects the optimal step to optimize the process.

2. The high-performance multi-axis machining automation control platform for parts according to claim 1 is characterized by: Record the strength and hardness of raw materials by referring to material manuals or standards; Arrange reasonable steps according to the functions and requirements of the parts by process engineers, or simulate the processing of the parts through CAD and CAE software, and record the number of processing steps of the parts; The part tolerance shall be based on the requirements of the part. If there is no clear tolerance requirement, refer to the general standards, check the relevant industry standards or technical requirements, and record the part tolerance.

3. The high-performance multi-axis machining automation control platform for parts according to claim 2 is characterized by: The material characteristic coefficient is calculated based on the hardness and strength of the raw material of the part, using the following formula: K m =α·H+β·σ, where: K m is the material characteristic coefficient, H is the hardness of the material, σ is the strength of the material, α and β are the weight coefficients of hardness and strength to processing difficulty; Set a weighting factor to include the number of processing steps and tolerance requirements into the criteria for judging whether a part is easy to process. The comprehensive processing difficulty coefficient is calculated using the following formula: Kprocess =γ·N+δ·T, where: K process is the processing difficulty coefficient, N is the number of processing steps, T is the tolerance requirement of the part, γ and δ are the weighting coefficients of the number of processing steps and tolerance requirements; Combining the material property coefficient and the processing difficulty coefficient, the comprehensive processing difficulty coefficient is expressed by the following formula: K total =λ·K m +(1-λ)·K process , where: K total is the comprehensive processing difficulty coefficient, and λ is a weighting coefficient representing the relative importance between material properties and processing difficulty; Set a threshold to judge the processing difficulty, set a threshold K threshold , judge according to the following rules: If K total ≤K threshold , the parts are easy to process and the processing difficulty is relatively low K total >K threshold , the parts are not easy to process and the processing difficulty is high.

4. The high-performance multi-axis machining automation control platform for parts according to claim 3 is characterized in that ； For parts that are not easy to process, the processing is broken down into more detailed steps, recorded as S1, S2, S3...S i ..., where i is the number of the step, S i is the i-th step; In the dependency matrix method of the processing process, a matrix is ​​constructed to represent the dependency relationship between the steps, where each matrix element A ij Indicates step S i and step S j If A ij is a non-zero element, it means step S j For step S i There is a direct dependency relationship. By counting the number of non-zero elements in the matrix, the size of the direct dependency of each step is further derived and normalized; For each step, the following information is recorded: operation type and operation parameters. The size of the state space is the number of combinations of all these operation parameters and state information. The operation type and parameters of each step can be selected independently, so the size of the state space of the step is the product of the number of operation types and the number of operation parameters, which is expressed as: state space = operation type × operation parameter.

5. The high-performance multi-axis machining automation control platform for parts according to claim 4 is characterized by: After receiving the correlation between steps and the state space, the algorithm selection module defines the correlation between steps and the state space as input variables and divides them into different fuzzy sets; The exploratory learning algorithm is defined as the output variable, which is divided into fuzzy sets; Formulate fuzzy rules to describe the connections between steps and the impact of state space definition on exploration learning algorithms; Carry out fuzzy reasoning based on fuzzy rules and determine the exploration learning algorithm; Using the determined exploration learning algorithm, between two step nodes, before proceeding to the next step, the probability of continuing to use the next step and the probability of using another step are calculated, and the probability of normally proceeding to the next step is marked as t. At this node, there is another 1-t probability of choosing another step. The probabilities of the two steps are compared to decide whether to proceed to the next step normally or choose another step. The decision is made based on the value of t. If t is higher, choose to continue using the next step, otherwise choose another step.

Images