Abstract
A test hammer is used to detect faulty components or slack bolts during impact inspection of marine diesel engine cranks. Since a prolonged work experience is necessary to acquire this technique, technology transfer is therefore difficult. Moreover, because this technology depends on the individual senses, the inspection accuracy is prone to vary among people. The authors have developed a system that can perform the above-described inspections and execute diagnostic procedures and thereby be more useful than human acoustic sense. We constructed an algorithm that estimates the fastening force of a crank-pin bolt at a connecting-rod large-end from sound, with which diagnostic tests were conducted. However, this algorithm is for a specific bolt only and is inapplicable to bolted joints of various types. This study conducted impact experiments using test pieces with varied component size and bolt tightening torque. First, the test piece size was estimated from impact sound data using discriminant analysis. That result was subjected to multiple regression analysis, a technique of multivariate analyses, from which the bolt tightening torque was estimated. The introduction of this two-step analytical method has provided us with a new tightening torque estimation algorithm for application to various bolted joints.
