In addition to the above constraints, practical considerations may dictate some upper and lower bounds on
the design variables. For example, the thickness of a shape may be limited by the capacity of the forming
equipment. Generally, the thickness of commonly used members varies from 0.5 mm to 6.0 mm.
3.4
NEURAL DYNAMICS OPTIMIZATION MODEL
Adeli and Park (1998) present a general neural dynamics model for optimization problems that guarantees a
stable and local optimum solution. This solved one of the fundamental problems of using ANNs for
optimization: How to find a neural dynamics system for a particular optimization problem that would
produce a stable and local optimum solution. The neural dynamics optimization model is robust and
particularly effective for large and complex optimization problems.
The constrained optimization problem (Eqs. 3.13.3), can be written in the following reduced form by
combining Eqs. (3.2) and (3.3):
Minimize
(3.20)
subject to inequality constraints
(3.21)
where J is the total number of inequality constraints including side constraints represented by Eq. (3.3).
Using the exterior penalty function method, the constrained optimization problem can be written as an
unconstrained optimization problem as follows:
(3.22)
(3.23)
(3.24)