Titanium alloy forging microscopic structure image recognition method based on machine learning
A technology of microstructure and machine learning, applied in the field of image recognition, can solve problems such as insufficient accuracy, low efficiency, and low efficiency of evaluation, and achieve the effect of ensuring classification accuracy, ensuring processing efficiency, and streamlining image data
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0035] In this embodiment, a method for recognizing the microstructure of titanium alloy forgings based on machine learning, such as figure 1 As shown, including the following steps:
[0036] Step 1. Obtain microstructure images of titanium alloy forgings of different materials, and classify the microstructure images of titanium alloy forgings based on the material performance level to form a data set;
[0037] Select the titanium alloy forgings corresponding to different material performance levels and obtain the corresponding microstructure images of the titanium alloy forgings, and store the microstructure images of the titanium alloy forgings according to the material performance levels. The material performance levels of the titanium alloy forgings can be divided into 1-11 A total of 11 levels, corresponding to 11 material performance levels, establish 11 material performance level folders, and then save the microstructure image of the titanium alloy forgings to the correspond...
Embodiment 2
[0045] This embodiment is further optimized on the basis of embodiment 1. The step 3 includes the following sub-steps:
[0046] Step 3.1: Segment the feature vector obtained in step 2 to obtain several feature vector segments;
[0047] According to step 2, the microstructure image of each titanium alloy forging can be expressed as a 324-dimensional feature vector. First, the 324-dimensional feature vector is segmented, and each 6-dimensional vector is regarded as a segment, that is, the 324-dimensional feature vector can be Divided into 54 feature vector segments.
[0048] Step 3.2: Perform pooling processing on the feature vector segment, select the one-dimensional vector with the largest value in the current feature limit segment, and use the one-dimensional vector to represent the current feature vector segment;
[0049] Then perform a pooling operation on the segmented feature vector segment, select the feature vector with the largest value in the vector segment as a one-dimension...
Embodiment 3
[0053] This embodiment is further optimized on the basis of the foregoing embodiment 1 or 2, and the step 4 includes the following sub-steps:
[0054] Step 4.1. Use the support vector machine to construct a corresponding number of classifiers corresponding to the material performance level, that is, construct 11 classifiers corresponding to the 11 material performance levels through the support vector machine;
[0055] Step 4.2, each classifier corresponds to a titanium alloy forging microstructure image of a material performance level to perform two classifications to obtain classifier models corresponding to different material performance levels;
[0056] Through the established classifier, the feature vector of the microstructure image of the titanium alloy forging is classified into two categories corresponding to the material performance level for classification training. The two classification means that a classifier classifies the microstructure image of the titanium alloy forg...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More