Design of an assembly planning system using unsupervised learning algorithm

An efficient approach using an unsupervised learning algorithm to generate assembly plans is proposed. Two algorithms, pattern clustering and retrieval algorithm (PCRA) and pattern adaptation algorithm (PAA), are presented, and are applied to a container assembly example. The symbolic knowledge slots adaptation is implemented in C language integrated production system (CLIPS). Assembly plans are encoded into patterns and fed into the designed self-organizing neural network. Based on the defined function, similar assembly plants automatically form a cluster. When a similar assembly plan is given, it can retrieve the appropriate cluster to identify the most approximate assembly pattern for adaptation