The study of measuring points on a plane curve by the method of gradually adding training case and concept of pattern assembly with fuzzy point

The purpose of this study is to investigate the determination of measuring points using a coordinate measuring machine for a plane curve in order to elevate the utilisation efficiency of measuring tasks. The study begins by establishing the data points to be measured on the design blueprint, and then presents the algorithm of determining the measuring points for a plane curve with the application of the neural networks. Hence, the learning model of back-propagation neural networks is applied in this article with the support of the training and testing of the network, and the allowed maximum perpendicular error to derive the proposed measuring points. To shorten the time for network training, the concept of pattern assembly is utilised to divide the curve into several segments based on the complexity of the curve shape. The coordinate points connecting the various segments are (Xn,Yn), where N = 1; 2; . . . These (Xn,Yn) points are fuzzy points, where N a 1; 2; . . . These (Xn,Yn) points are fuzzy points. It has a middle transition area around the fuzzy point. If a measuring point locates at this middle transition area, we propose that using the fuzzy set method and considering the shorter measuring time to decide this measuring point locates which measuring segment.