The majority of engineering designs can be described as configuration design. The solution of configuration design is not only the satisfaction of constraints and requirements, but the solution requires the optimized variant. However, there are usually multiple objectives and corresponding design problem in a multiobjective one. Even for relatively small problems the solution space is large and the computational complexity can be very high with exponential growth. Another problem of configuration design is the usual nonlinear and non-monotonic character of engineering design. The efficient computation of multiobjective global optimisation is a powerful general way for solving the nonlinearity and non-monotonicity of design problems.
The paper deals with two issues: (1) the description of general approaches how to reduce this computational complexity and thus to realize tractable treatment of such design and (2) the description of application of this approach for design of parallel kinematics machines. The described general approaches for the reduction of computational complexity are the decomposition of design space into hierarchy of subspaces, the development of computational tools for global evaluation of mechanical properties and usage of global optimization tools that enable efficiently to find admissible solutions and to reconstruct the Pareto sets.
This general approach has been successfully applied for design methodology of parallel kinematics machines. Parallel kinematics machine means a platform with actuator (gripper for manipulation in robotics, cutting tool for machining in machine tools) that is suspended on several kinematic chains in parallel. The developed design methodology has been successfully used for the industrial design of the new horizontal machine tool TRIJOINT 900H with significantly improved mechanical properties.