基于MATLAB图像处理的铸造表面粗糙度测量方法

Measuring Method of Surface Roughness of Casting Based on MATLAB Image Processing

采用直接测量法测量铸造表面粗糙度时,不仅操作繁琐,而且仪器探头容易损坏。本文提出了一种基于BP神经网络模型为核心的模拟训练辨识方法,建立了图像特征值与粗糙度的训练辨识关系。以铸铁试件为实验样本,利用MATLAB对图像进行算法分析,得出占空比和亮度比两个特征值,推出特征值规律;利用MATLAB中的BP神经网络模拟系统与粗糙度参数Ra进行对比。结果表明,本方法试验结果较理想,211次测试的粗糙度参数训练精度高达99.98%。

Aiming at the problems of cumbersome operation and easy damage of instrument probe when measuring the surface roughness of castings using direct measurement method, a simulation training identification method based on BP neural network model was proposed, and the training identification relationship between image eigenvalue and roughness was established. Taking nine cast iron test pieces as the experimental samples in this work, their surface images were analyzed by MATLAB so that two eigenvalues(duty cycle and brightness ratio) were obtained and then the rule of eigenvalue was deduced. The comparison between the simulation system of BP neural network in MATLAB and the roughness parameter Ra was made. The results showed that the training accuracy of the proposed method was as high as 99.98% via 211 times of testing training, with effective and satisfactory results.