A neural network approach to determining optimal inspection sampling size for CMM

Automated Coordinate Measuring Machine Inspection Planning Knowledge Capture and Formalization

“…In the effort to support the time-consuming task of programming a CMM, many researchers attempted to design and build semi-or fully-automated inspection planning systems. Some of these focus only on finding optimal solutions for some aspects of inspection planning, such as the determination of inspection probe accessible directions [5][6][7][8], part orientations and setup [9][10][11], sampling methods for feature probing and reconstruction [12][13][14][15][16] and path generation [17][18][19][20], while other works attempt to address and provide complete solutions for the whole measurement strategy. The main approaches used [2] covering the widest range of systems researched are expert systems and virtual reality (VR) based systems.…”

Selection of sampling points for accurate evaluation of flatness error using coordinate measuring machine

“…The sampling strategy comprises of the optimum sample size and the locations of the sampling points. The determination of the sample size is a complicated process since it is affected by many factors such as the manufacturing process used, tolerance specifications, error evaluation method and confidence level of measured results (Zhang et al, 1996). Since the sample should be a good representative of the entire surface, the location of the points should be such that a maximum amount of information is obtained.…”

“…These reports have not discussed the measurement uncertainty arising while considering the sampling size. Zhang et al (1996) have used a neural network approach to determine the sample size for the inspection of holes using the process type, size of hole and tolerance band as factors. Hwang et al (2002) have used a hybrid neuro-fuzzy approach considering the tolerance and geometry features as factors.…”

Use of genetic algorithm to design optimal neural network structure

“…Several repeated solutions with different initial weights and network parameters are used to converge the optimal solution. There is no general framework to select the optimum NN architecture and its parameters (Chung and Kusiak, 1994;Goh, 1995;Kusiak and Lee, 1996;Yoon et al, 1993;Zhang et al, 1996). Although some recent research work has contributed to Use of genetic algorithm determine the number of hidden layers, the number of neurons in each layer and selecting the learning rate parameters, the results are still not at satisfactory level to be accepted as general rules for generating optimal NN architecture.…”