A constraint-based inference system for satisfying design constraints

We propose an efficient algorithm for the purpose of satisfying a wide rang e of design constraints represented with equality and inequality equations as well as production rules. The algorithm employs simulated-annealing and a production rule inference engine and works on design constraints represen ted with networks. The algorithm fulfills equality constraints through cons traint satisfaction processes like variable elimination while taking into a ccount inequality constraints and inferring production rules. It can also r educe the load of the optimization procedure if necessary. We demonstrate t he implementation of the algorithm with the result on machine tool design.