The Kinematics and Dynamics of a Dextrous Parallel Manipulator for Structural Inspection Applications

摘要

Traditional NDT methodologies rely on the use of non-invasive sensors for the detection of structural inclusions, cracks, delaminations, and flaws, and full geometrical coverage can be a considerable challenge in the case of structures with complex geometries. New sensing technologies based on the generation and subsequent measurement and analysis of acoustic wave propagation within structures by means of sophisticated algorithms can reveal considerable information from a relatively low number of low-cost sensor-actuators fitted semi-permanently (or possibly permanently) to the structure. In addition to this more traditional NDT sensing systems will continue to play an active role, particularly in very large mechanical and civil structures where it is technically or economically unfeasible to fit, or retro-fit, integrated piezo sensor-actuators and local computing hardware. In such cases roving NDT sensors will continue to be used for periodic maintenance and safety testing. Accurate sensor positioning depends on the use of highly dextrous manipulator technology and this paper considers the mechanics of a parallel 2-link manipulator in such a way that an accurate reduced order nonlinear analytical model is obtained for the dynamics. Most parallel manipulator models are either somewhat trivialised reduced order linear models or are based on numerical codes such as ADAMS or DADS, from which explicit equations of motion are not available. Whilst models built using codes such as these will yield accurate results for specific data cases they are not generic and cannot, therefore, lead to global analytical solutions. This paper summarises the principal steps involved in defining the three dimensional geometry of a feasible 2-link parallel manipulator and from that obtains certain kinematical relations from which the dynamical equations of motion are derived. It is shown that this leads to a reduced order nonlinear analytical model which can then either be solved for specific operational data or prepared for approximate analytical solution. An experimental system will be built at the University of Glasgow in the near future in order to validate the modelling work discussed in this paper.