Current software technologies enable modeling of the open ended nature of the design process in a modular and efficient manner using object oriented and other such methods. Various AI techniques have enabled representation of knowledge in visible and easily up gradable fashion for use in such design software. All these technologies have rendered design software more flexible and user friendly for solving the open ended design problem emulating usual design practice. However, a designer learns from his earlier design experiences and thus improves his design capability and design output with time. Although machine learning technologies have made great strides in the past decade, demonstration of the ability to capture the design knowledge in real time, assimilating it and using it subsequent designs has been very limited. In this paper the Locally Weighted Projection Regression method is briefly explained and its application to online incremental learning of structural engineering design is demonstrated.
展开▼
