Soft computing system for motion control

The novel concept of pseudo-errors is proposed for self-organizing neuro-fuzzy system (SO-NFS) in this paper. The SO-NFS is viewed as a soft computing system. Based on the concept of pseudo-errors and a clustering algorithm, a neuro-fuzzy system possessing self-organizing ability and learning ability is presented. Pseudo-errors are viewed as the prior knowledge of a given unknown plant, and they are used as important information for the SO-NFS to self-organize its initial knowledge base. Pseudo-errors are potential errors that could be occurred in application. They can be used to indicate where the fuzzy regions are of interest in the input space for application, so to avoid the waste of computational resource. With the information of pseudo-errors, a more concise structure of neuro-fuzzy system can be self-organized. The well-known random optimization (RO) algorithm is used to search for a near-optimal set of parameters for the SO-NFS. The proposed approach is applied to motion control of an auto-warehousing crane system.