Statistical Modelling of Nominal and Measured Mechanical Surfaces

Abstract

Modern manufacturing processes require unambiguous description of product morphology. In spite of numerous successes in the development of new mathematical tools, there not exists a method providing complete and coherent information on the product shape along its lifecycle. Consequently, industrial methods currently employed in dimensional and geometrical controls do not fully satisfy designers, manufactures and customers. A possible solution could be a statistical description of product shape because it has strong mathematical basis, uses powerful analysis tools and provides a single unifying model along the product development process. The comparison with industrial practice and deterministic mathematical tools in the design, manufacturing and verification phases, shows some interesting advantages of the probabilistic approach. The paper illustrates the theoretical basis of the probabilistic approach, provides the instruments necessary to its implementation and, finally, shows some applications in the inspection of mechanical objects.