Probe automatic configuration method of CMM detection system
By using an automated configuration method and employing heuristic and intelligent optimization algorithms, the configuration of the probe and extension rod for the CMM inspection system is automatically generated. This solves the problems of long manual configuration time and errors, achieving an efficient and accurate measurement scheme and improving the automation and inspection quality of the CMM inspection system.
Patent Information
- Application Number
- CN202410562806.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-08
- Publication Date
- 2024-11-12
AI Technical Summary
In existing CMM inspection systems, probe configuration mainly relies on manual labor, which leads to long configuration time, easy errors, low efficiency when measuring complex parts, and increased labor costs.
An automatic configuration method is adopted, which imports part models and measurement point information, and uses heuristic and intelligent optimization algorithms to automatically generate configuration schemes for probes and extension rods. Combined with collision detection and iterative optimization, the configuration with the fewest probes is achieved.
It improves the efficiency and accuracy of probe configuration, reduces the number of probes and extension rods, lowers assembly time and labor costs, and enhances the automation level and inspection quality of the CMM inspection system.
Smart Images

Figure CN118939902A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method for automatically configuring a measuring needle of a CMM detection system, and mainly relates to a method for automatically planning the configuration of a measuring needle suitable for the CMM detection system. Background Art
[0002] The dimensional measurement of processed products is a key link in achieving product quality monitoring, control and improvement. Five-axis coordinate measuring machines (CMMs) are widely used in aerospace, automobile body and other fields due to their advantages such as high detection accuracy, high detection reliability and low long-term cost. For processed products with high processing accuracy, multiple probe configurations are often required to complete the measurement of the product. However, the allocation of probes, extension rods and five-way adapters in the workstation affects the efficiency of the entire measurement. At present, the configuration of probes in CMM measurement is generally based on manual configuration, which has the following disadvantages: 1) The configuration scheme often requires manual configuration, resulting in a long time for scheme configuration; 2) For complex parts, there are many dimensions to be measured and many special-shaped angles, which is prone to errors in probe configuration. These disadvantages undoubtedly increase the time and labor cost of product measurement planning. Based on the above disadvantages, we hope to develop an automatic probe configuration method for CMM detection system. It aims to improve the overall program generation efficiency by automatically generating CMM detection system schemes and associating CMM automatic programming. Summary of the invention
[0003] The purpose of the present invention is to provide a method for automatically configuring the stylus of a CMM detection system, which can automatically generate the required stylus and matching extension rod through programming according to the part model and measuring point information, so as to minimize the number of stylus configurations while meeting the measurement accuracy requirements.
[0004] In order to achieve the above effect, the present invention comprises the following steps:
[0005] S101, importing the model and the measuring point information, and classifying the points to be measured according to the vector direction according to the adapter plane according to the position and vector direction of the measured features in the measuring point information;
[0006] S102. Due to the measurement specification constraints of the stylus, some measuring points in space may exist on multiple planes at the same time, and it is necessary to map such points to the planes to which they belong;
[0007] S103, classifying the measuring points according to the surface, forming a mapping from the measuring points to the surface, seeking the measuring needle configuration according to the surface, and simplifying the process of measuring point to measuring needle configuration;
[0008] S104, simplifying the problem of allocating measuring point vector information into the problem of matching measuring points in each adapter plane with the measuring needle and the extension rod; performing collision detection on different measuring needles, extension rods and measuring point information configurations to form multi-dimensional interference information, and finding a suitable configuration solution set according to the interference matrix; allocating measuring needle configurations to the measuring points in turn according to the interference matrix to meet the optimization goal of having the least number of measuring needle configuration types;
[0009] S105, based on the interference matrix obtained in S104 and the constraints on the stylus configuration types, converting the matching problem between the measuring points in the plane of each adapter and the stylus and the extension rod into a configuration with the least number of stylus configuration types;
[0010] S106, using an intelligent optimization algorithm, a neural network or a machine learning method to solve the configuration with the least number of probe configuration types, and obtain an optimal configuration solution;
[0011] S107. The best match between the measuring point, the measuring needle and the extension rod in this plane is solved through continuous iteration.
[0012] S108, return to S103, redistribute the measuring points and the conversion head plane through the heuristic algorithm, repeat S104-S107 until the iteration condition is met, and output the probe configuration result.
[0013] Furthermore, in step S101, the type of the plane is determined by the five-direction adapter, and the adapter plane to which the point belongs is determined based on the angular relationship between the vector information of the measuring point and the adapter plane.
[0014] Furthermore, in step S103, a heuristic algorithm, a neural network or a machine learning method is adopted to classify the measuring points by surface, form a mapping from measuring points to surfaces, and seek the configuration of the measuring pins according to the surfaces, thereby simplifying the process of configuring the measuring points to the measuring pins in the original problem.
[0015] Furthermore, in step S104, the interference matrix indicates whether a combination of a probe and an extension rod collides at a certain measuring point.
[0016] Furthermore, in step S107, the evaluation index of the iteration condition is:
[0017]
[0018] in, Represents a collection of measuring point configurations, such as {[surface, measuring needle, extension rod]}; calculate The number of configuration types that do not interfere with each other, and the minimum value num is found through argmin().
[0019] Furthermore, in step S108, the measuring points and the conversion head plane are redistributed by using a heuristic algorithm, a neural network or machine learning, and S104-S107 are repeated, and the heuristic algorithm or neural network is used to continuously solve until the iteration condition is met to obtain the result of the probe configuration.
[0020] Compared with the prior art, the advantages of the present invention are as follows: the present invention classifies the measuring points according to the conversion head plane through a heuristic algorithm, and this classification makes it easy to include more measuring points in the same conversion head plane, so that the configuration categories of the measuring needle and the extension rod are fewer; then the corresponding measuring needle combination is performed for the measuring points in this plane, and the best matching of the measuring points, the measuring needle and the extension rod in this plane is solved through continuous iteration by an intelligent optimization algorithm, a neural network or a machine learning method. Then the heuristic algorithm is used to return the allocation of the measuring points and the planes to which they belong, and the planes to which the measuring points belong are updated, and the iteration is repeated until the optimal measuring needle configuration is obtained.
[0021] Through the above method, while ensuring the CMM measurement accuracy, on the one hand, we can quickly obtain the probe configuration, and on the other hand, the number of probes and connecting rods is reduced, which reduces the assembly and unloading of accessories, reduces the time spent on the configuration plan, and improves the efficiency of actual manufacturing. It improves the automation of the CMM detection system and ensures the reliability of the detection quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 A flow chart of a method for automatically configuring a stylus for a CMM inspection system.
[0023] Figure 2 Graph of the genetic algorithm encoding process. DETAILED DESCRIPTION
[0024] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further described below:
[0025] The present invention proposes a method for automatically configuring a probe of a CMM detection system, such as Figure 1 As shown, the following steps are included:
[0026] S101, import the model and the measuring point information. According to the position and vector direction of the measured feature in the measuring point information, the points to be measured are classified according to the vector direction and the adapter plane. The type of plane is determined by the five-direction adapter. According to the angle relationship between the vector information of the measuring point and the adapter plane, the adapter plane to which the point belongs is determined;
[0027] S102. Based on the method proposed in S101, due to the measurement specification constraints of the stylus, some measuring points in space may exist on multiple planes at the same time. We need to map such points to the planes to which they belong.
[0028] S103. Based on the problem raised in S102, it is necessary to ensure the mapping between the measuring points and the planes to which they belong. A heuristic algorithm is used to classify the measuring points by plane to form a mapping from the measuring points to the planes. For example, the ε-greedy algorithm is used. In addition to the ε-greedy algorithm, other heuristic algorithms, neural networks, machine learning, etc. can achieve this function and are also within the scope of protection. Therefore, we use this type of algorithm to classify the measuring points by plane, and seek the configuration of the measuring needle according to the plane, which simplifies the process of configuring the measuring points to the measuring needle in the original problem;
[0029] S104. Based on the method proposed in S103, we simplify the original problem of allocating measuring point vector information into the problem of matching measuring points with measuring probes and extension rods in each adapter plane. We perform collision detection on different measuring probes, extension rods and measuring point information configurations to form a three-dimensional interference matrix, and find a suitable configuration solution set based on this interference matrix. In addition, we use this interference matrix to allocate measuring point measuring probe configurations in turn to meet the optimization goal of having the least number of measuring probe configuration types.
[0030] S105. Based on the interference matrix obtained in S104 and the constraints on the probe configuration types, we transform this problem into a solution with the goal of finding the configuration with the least number of probe configuration types;
[0031] S106. For the problem raised in S105, we can use intelligent optimization algorithms or neural networks to solve this problem, such as using genetic algorithms to solve it, and finally get the optimal configuration solution. Take genetic algorithms as an example to explain. Other algorithms that can achieve this function are within the scope of protection. Figure 2 It can be seen that we use the configuration type of the adapter surface as the gene of the genetic algorithm, and encode the probe configuration of the adapter plane in turn. The encoding process gives priority to the configuration type that has already appeared under the premise of ensuring non-interference. And take the adaptability of this gene to the adapter plane as an evaluation index, randomly select and cross from genes with higher adaptability, and then create a new excellent gene pool, and take out from the excellent gene pool in the form of probability to replace the original gene for mutation, so as to promote the population to optimize in the direction of fewer configuration types;
[0032] S107. Solve this problem based on the intelligent optimization algorithm, neural network or machine learning algorithm proposed in S106, such as genetic algorithm, and solve the best match between the measuring point and the measuring needle and the extension rod in this plane through continuous iteration. In the solving process, under the premise of ensuring non-interference, the configuration types that have appeared are given priority;
[0033] S108, return to S103, redistribute the measuring points and the conversion head plane by using a heuristic algorithm, repeat S104-S107 until the iteration condition is met, and output the probe configuration result;
[0034] Specifically, in step S101, whether the measuring point is on the corresponding plane is determined by determining whether the angle between the measuring point vector direction and the external normal of the adapter plane is less than 60 degrees.
[0035] In step S102, since the measuring point may belong to multiple planes of the conversion head at the same time, it is proposed to map the measuring point to one plane of the conversion head.
[0036] Step S103 allocates as many measurement points as possible to the adapter plane through a heuristic algorithm to obtain a mapping from the point to the adapter plane. Taking the ε-greedy algorithm as an example, each time there is an ε probability of exploration, there is a (1-ε) probability of using the learned data, as shown in the formula:
[0037]
[0038] where Q occurs with probability ε ε Event, occurs with probability (1-ε) Q 1-ε Event. Therefore, we can use the ε-greedy algorithm to configure the measuring points for the plane with diversity. We use this kind of heuristic algorithm to classify the measuring points by surface, form a mapping from measuring points to surfaces, and seek the configuration of the measuring needle according to the surface, which simplifies the process of configuring the measuring point to the measuring needle in the original problem. In addition to the ε-greedy algorithm, other heuristic algorithms, neural networks, machine learning, etc. can achieve this function and are also within the scope of protection of the rights;
[0039] Step S104 simplifies the original planning of the measuring point vector information into the planning of the corresponding adapter plane. By performing collision detection on different probes, extension rods and measuring point information configurations, a corresponding three-dimensional interference matrix is formed. The interference matrix indicates whether a combination of a probe and an extension rod collides at a certain measuring point. We need to find a suitable configuration solution set through this interference matrix.
[0040] In step S105, we use the interference matrix to allocate the probe configurations to the measuring points at one time to meet the optimization goal of having the least number of probe configuration types.
[0041] Step S106 converts the original problem of finding the minimum number of probe configuration types into the problem of finding the configuration with the minimum number of probe configuration types, and uses intelligent optimization algorithms, neural networks or machine learning to solve this problem. Through continuous iteration, the best match between the measuring point and the probe and the extension rod in this plane is solved. This is explained using the genetic algorithm as an example. Other algorithms that can achieve this function are within the scope of protection. Figure 2It can be seen that we use the configuration type of the adapter surface as the gene of the genetic algorithm, and encode the probe configuration of the adapter plane in turn. The encoding process gives priority to the configuration type that has already appeared under the premise of ensuring non-interference. And take the adaptability of this gene to the adapter plane as an evaluation index, randomly select and cross from genes with higher adaptability, and then create a new excellent gene pool, and take out from the excellent gene pool in the form of probability to replace the original gene for mutation, so as to promote the population to optimize in the direction of fewer configuration types;
[0042] In step S107, under the premise of ensuring that the interference matrix does not interfere, an intelligent optimization algorithm, neural network or machine learning is used to make the best configuration plan for the allocation problem of the measuring points, the measuring needles and the extension rods in this plane, so as to obtain the best matching of the measuring points, the measuring needles and the extension rods in this plane. The evaluation index we use is shown in the formula:
[0043]
[0044] in, Represents a collection of measuring point configurations, such as {[face, probe, extension rod]}. calculate The number of configuration types that do not interfere with each other, and the minimum value num is found through argmin().
[0045] Step S108 reallocates the measuring points and the conversion head plane by using a heuristic algorithm, a neural network or machine learning, and repeats S104-S107, using a heuristic algorithm or a neural network to continuously solve until the iteration condition is met, and the result of the probe configuration is obtained.
Claims
1. A method for automatically configuring a stylus of a CMM detection system, characterized in that: The following steps are involved: S101, importing the model and the measuring point information, and classifying the points to be measured according to the vector direction according to the adapter plane according to the position and vector direction of the measured features in the measuring point information; S102. Due to the measurement specification constraints of the stylus, some measuring points in space may exist on multiple planes at the same time, and it is necessary to map such points to the planes to which they belong; S103, classifying the measuring points according to the surface, forming a mapping from the measuring points to the surface, seeking the measuring needle configuration according to the surface, and simplifying the process of measuring point to measuring needle configuration; S104, simplifying the problem of allocating measuring point vector information into the problem of matching measuring points in each adapter plane with the measuring needle and the extension rod; performing collision detection on different measuring needles, extension rods and measuring point information configurations to form multi-dimensional interference information, and finding a suitable configuration solution set according to the interference matrix; allocating measuring needle configurations to the measuring points in turn according to the interference matrix to meet the optimization goal of having the least number of measuring needle configuration types; S105, based on the interference matrix obtained in S104 and the constraints on the stylus configuration types, converting the matching problem between the measuring points in the plane of each adapter and the stylus and the extension rod into a configuration with the least number of stylus configuration types; S106, using an intelligent optimization algorithm, a neural network or a machine learning method to solve the configuration with the least number of probe configuration types, and obtain an optimal configuration solution; S107. The best match between the measuring point, the measuring needle and the extension rod in this plane is solved through continuous iteration. S108, return to S103, redistribute the measuring points and the conversion head plane through the heuristic algorithm, repeat S104-S107 until the iteration condition is met, and output the probe configuration result.
2. The automatic stylus configuration method of a CMM detection system according to claim 1, characterized in that: In step S101, the type of the plane is determined by the five-direction adapter, and the adapter plane to which the point belongs is determined based on the angular relationship between the vector information of the measuring point and the adapter plane.
3. The automatic stylus configuration method of a CMM detection system according to claim 1, characterized in that: In step S103, a heuristic algorithm, a neural network or a machine learning method is adopted to classify the measuring points by surface, form a mapping from measuring points to surfaces, and seek the configuration of the measuring needle according to the surface, thereby simplifying the process of configuring the measuring points to the measuring needle in the original problem.
4. The automatic stylus configuration method for a CMM detection system according to claim 1, characterized in that: In step S104, the interference matrix indicates whether a combination of a probe and an extension rod collides at a certain measuring point.
5. The automatic stylus configuration method for a CMM detection system according to claim 1, characterized in that: In step S107, the evaluation index of the iteration condition is: in, Represents a collection of measuring point configurations, such as {[surface, measuring needle, extension rod]}; calculate The number of configuration types that do not interfere with each other, and the minimum value num is found through argmin().
6. The automatic stylus configuration method for a CMM detection system according to claim 1, characterized in that: In step S108, the measuring points and the conversion head plane are reallocated by heuristic algorithms, neural networks or machine learning, and S104-S107 are repeated, and the heuristic algorithms or neural networks are used to continuously solve until the iteration conditions are met to obtain the result of the probe configuration.
