Design By Expectation: a framework for engineering design optimization

An engineering design often requires sophisticated understanding of the domain knowledge and the designer's creativity. To optimize the engineering design, one needs to explore a combinatorial search space. We propose a framework, called Design By Expectation (DBE), which provides a collaborative scheme for genetic algorithms (GA's) and domain-specific knowledge to carry out the engineering design optimization. DBE exploits the adaptation of GA's in approaching better solutions, which compensates for the weak creativity of computers. We make the DBE framework more flexible by developing a set of expectation interpreters that release the strict requirements on the formalities of objectives and constraints required in traditional optimization techniques. Finally, we use two practical examples: power system stabilizer and streetlight placement, to demonstrate the applicability of the DBE framework.<>